fix(typecheck,compile): allow deferred parameter [:] dims and speed u…

[yeus]

Summary

• Fixes a compile correctness gap: parameter/constant [:] dimensions can now remain
  unresolved until later binding/instantiation (MLS-aligned), instead of failing early in
  typecheck.
• Speeds up compile-heavy flows by optimizing clock timing precompute and strict path
  matching.
• Adds compile/check timing metadata in source-root/WASM compile APIs for better
  diagnostics.
• Adds a profiling utility script for reproducible perf/flamegraph and AI-readable profiling
  artifacts.
• Measured impact on the target model
  WindPowerPlants.Components.GenericVariableSpeedGeneratorElectrical: compile-check wall
  time reduced from ~1m42s to ~12.7s (~8x faster).

Code Size Budget (required)

• production_lines_added: 455
• production_lines_deleted: 44
• test_lines_added: 26
• test_lines_deleted: 0
• public_items_added: 12
• public_items_removed: 0
• files_touched: 10
• net_added_lines: 437

If net_added_lines is positive, add:

• Why this net growth is required.
  • The growth is primarily from two necessary additions: (1) timing instrumentation
    surfaces in compile APIs and (2) a reusable profiling script (scripts/
    profile_rumoca_model.sh). The performance fixes in clock precompute also require new
    indexing/data-structure code to avoid repeated scans.
• Which code was removed/merged as part of the first compression pass.
  • Replaced repeated reverse-alias full scans with a precomputed reverse index in clock
    precompute.
  • Reused allocation/scratch paths instead of repeated per-iteration structures in hot
    code.
  • Replaced repeated path canonicalization calls with cached canonicalized keys in strict
    diagnostics.
• Follow-up cleanup ticket/commit for remaining growth (if any).
  • Follow-up: collapse duplicated timing JSON construction between bind-wasm entry points
    into shared helpers and tighten script option parsing utilities (planned as a cleanup
    commit after merge).

Design Notes

• Why each new abstraction was necessary.
  • StrictCheckTiming: needed to expose strict-check phase latency without mixing concerns
    into error payload formatting.
  • SourceMap reverse-alias index in clock precompute: needed to remove O(N) scans per
    lookup in hot inference loops.
  • Path canonicalization cache: needed to avoid repeated filesystem canonicalize calls in
    strict failure collection.
• Why this was not solved by editing an existing module directly.
  • The bottlenecks crossed module boundaries: typecheck validation, DAE clock precompute,
    compile session strict-check path, and wasm/source-root API serialization. A single-
    module tweak would not address end-to-end compile latency and observability.
• What duplicate/legacy paths were deleted.
  • Legacy repeated reverse-alias lookup behavior in clock inference was effectively
    replaced by indexed lookup.
  • Repeated ad-hoc canonicalization in strict path equality was replaced by centralized
    cached resolution.

Testing

• Key command(s) run.
  • cargo fmt --all
  • cargo check -p rumoca-phase-dae -p rumoca-compile
  • ./target/release/rumoca check ... --model
    WindPowerPlants.Components.GenericVariableSpeedGeneratorElectrical ... (before/after
    timing comparison)
  • Workspace integration run (compile-only):
    • yarn modelica:compare run --compile-only --compile-debug --library-zip public/
      modelica-libraries/WindPowerPlants.zip --targets-prefix WindPowerPlants --target-
      class-types model,block --json
• Files changed covered by tests.
  • crates/rumoca-phase-typecheck/src/typechecker/late_methods.rs is covered by new
    regression test in crates/rumoca-phase-typecheck/src/tests.rs
    (test_parameter_colon_dimension_without_binding_is_allowed).
  • Performance-related paths were validated by compile-only integration runs and direct
    model compile checks.
• Updated a brittle WASM alias test to ignore only non-semantic timing metadata fields
  (__compile_phase_timing, __compile_check_timing) before comparing compile vs compile_to_json,
  while keeping strict equality for the full semantic payload.

Reviewer Checklist

• Size budget section completed.
• Positive net diff has explicit compression justification.
• New APIs are required and minimal.
• Old/new parallel paths were removed unless explicitly migrating.

[jgoppert]

I like this script, but it would be good if we could leverage the rum developer tool for things like this. So far I've managed to avoid bash/python scripts entirely, which helps with crossplatform support.