feat(scheduler-core): implement REPL breakpoints and state inspection (task 12.2)

Adds the interactive simulation REPL's breakpoint and state-inspection
commands (Req 5 AC 22): `break <event_type> [key=value ...]` / `info
break` / `delete <id>|all` on event and decision types, and `info
session/job/queue/worker/resource/summary`.

repl.c (370 lines after task 12.1) couldn't hold this task's full
surface under the project's 800-line cap, so it's split three ways
(matching the snapshot.c/snapshot_wire.c/snapshot_load.c precedent):
repl.c (execution control + dispatch), repl_break.c (matching/CRUD +
the three new driver-facing hooks), repl_info.c (info dumps), sharing
state via repl_internal.h. sim_loop.c gained 3 new bare hook calls
(repl.h inline wrappers, same pattern as task 12.1's two) at the
points design.md's pseudocode requires -- per decoded event before it
joins the batch, per session boundary, and per drained decision --
landing sim_loop.c at exactly 800/800 lines.

Two real bugs found and fixed during development (bug-1506, bug-1507):
an uninitialized xstr_table left breakpoint filter fields silently
NULL instead of failing loud, and repl_breakpoint_free() called a bare
free() on an RTTI-refcounted makeString() string, corrupting the heap.

Review found one gap: `info job` printed only a resreq presence flag
instead of the actual resreq text, when format_resreq() already
existed in this repo for exactly this purpose. Fixed to match.

Scheduler suite: 35/35 clean (was 34/34). RED-phase evidence (2 real
bugs plus 2 synthetic probes, all reverted with confirmed zero
residue) recorded in tasks.md.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JPPXeKSUXUqcqvzNXY1rzJ

.kiro/specs/scheduler-core/tasks.md

@@ -362,11 +362,12 @@
   - _Depends: 11.2_
   - _Verified: RED-phase reproduction performed during review round 2 — temporarily changed `repl_should_pause()` (repl.c) from `return (rs->step_budget <= 0);` to `return false;`, rebuilt, ran `sched_repl_execution_control_test`: 2/4 tests failed exactly as expected (`test_should_pause_fires_on_exactly_the_nth_session`, `test_step_1_processes_exactly_one_session_then_quit_exits`), the 2 unrelated tests (`test_should_pause_never_fires_in_continue_mode`, `test_pre_replay_pause_blocks_before_any_event_is_processed`) stayed green. Reverted; rebuilt; full scheduler suite 34/34 GREEN._
 
-- [ ] 12.2 Implement REPL breakpoints and state inspection
+- [x] 12.2 Implement REPL breakpoints and state inspection
   - Support setting/listing/deleting breakpoints on event type with optional field filters, and inspection commands reporting current simulation position, and job/queue/worker/resource/summary state
   - Observable: setting a breakpoint on a specific event type and field value pauses the REPL exactly when a matching event is about to be processed; each inspection command reports state matching what a snapshot dump of the same moment would show
   - _Requirements: 5.22_
   - _Depends: 12.1_
+  - _Verified: 2026-08-07 — repl.c (task 12.1, 370 lines) split three ways under this project's 800-line cap (task 12.2's own feature surface did not fit in one file): repl.c (416 lines, execution control + dispatch, unchanged in scope), repl_break.c (breakpoint kind/filter matching, breakpoint CRUD, and the three new driver-facing hooks xce_repl_on_event()/xce_repl_on_begin_round()/xce_repl_on_decision()), repl_info.c (`info job`/`queue`/`worker`/`resource`/`summary`/`session`/`break` dumps) -- all sharing private state via a new repl_internal.h (the full xce_sched_repl_state_s definition, moved out of repl.c). sim_loop.c gained exactly 3 new bare call expressions (xce_repl_check_event()/xce_repl_check_begin_round()/xce_repl_check_decision(), repl.h inline wrappers gated on sim->interactive, same pattern as task 12.1's two existing hooks) at the three points design.md's pseudocode requires: per decoded ReplayEvent before it joins the accumulated batch, per ReplaySession boundary before the batch is pushed, and per drained decision after sim_print_decision() -- landed sim_loop.c at exactly 800/800 lines (trimmed its own header comment to make room). repl_should_pause()'s stale doc comment (which wrongly predicted this task would extend it) was corrected: breakpoints are a separate mechanism, never folded into the `step N` pause path. test_sim_isolation.c (task 11.3, already extended by 12.1) extended again to scan repl_break.c/repl_info.c/repl_internal.h (8-file precedent). New test_repl_breakpoints_inspection.c (17 tests, 4 groups: command parsing/listing, pure matcher predicates, state-inspection dumps against a real xce_sched_core_t, full-integration scripted-stdin pause proof for both an event breakpoint and a decision breakpoint). RED-phase: two REAL bugs surfaced and fixed during development (bug-1506: test fixtures calling repl_dispatch()'s `break` command without first calling xce_sched_core_init() left makeString()'s backing xstr_table uninitialized, silently returning NULL breakpoint fields instead of failing loud; bug-1507: repl_breakpoint_free() called a bare free() on a makeString()-produced xstring -- RTTI reference-counted, not a plain malloc() chunk -- causing a heap-corruption abort, fixed by never freeing those fields at all, matching kill_sched_job()/kill_sched_queue()'s existing "makeString()'d fields live for the process's lifetime" precedent). PLUS two targeted synthetic RED-phase probes reproduced independently after both real bugs were fixed: (1) repl_find_event_breakpoint() hardcoded to always return NULL -- rebuilt, ran the suite, exactly 3 of 17 tests failed (test_event_breakpoint_matches_jobid_filter, test_event_breakpoint_matches_state_filter, test_event_breakpoint_pauses_repl_at_matching_event), all others (including decision/begin_round matching and every info dump) stayed green; (2) repl_job_state_name() hardcoded to return a fixed bogus string -- exactly 1 of 17 failed (test_info_job_reports_state_queue_priority). Both probes reverted (grep-confirmed zero residue), rebuilt, reconfirmed green. Final: 35/35 scheduler suite (was 34/34), whole-project build clean. Remediation (2026-08-07): review found `repl_info_job()`'s resreq= field reported only a `set`/`(none)` presence flag instead of real content, despite `format_resreq()` (resreq.h) already existing and already used one file away by `build_wire_job()` (snapshot_wire.c) for this exact purpose; fixed by calling `format_resreq()` in `repl_info_job()` (mirroring `build_wire_job()`'s WARN-and-fallback error handling, freed after printing since this dump is transient), corrected the function's stale doc-comment claim that no print-form existed, and extended `test_info_job_reports_state_queue_priority` to assert the real formatted text (`"Primary:"`/`"Tasks: min = 0, max = 0"`/`"Rusages:"`) appears instead of the old boolean flag. Still 35/35 scheduler suite.
 
 - [ ] 12.3 Implement REPL snapshot capture and simulation parameter commands
   - Support writing a full snapshot to a file at the current simulation position, and adjusting speed/output-file/verify-mode from the prompt without restarting