Add reference graphs for scientific and native-extension languages

[Widthdom]

Summary

Several scientific/native-extension languages expose symbols but no authoritative references/graphs, and ambiguous .m files still expose no structured capability at the ambiguity surface.

Integrated v1.39.4 findings

• Add reference and graph extraction for Nim
• Add reference and graph extraction for MATLAB
• Add reference and graph extraction for Julia
• Add reference and graph extraction for D
• Add reference and graph extraction for Cython
• Add structured symbol and reference support for ambiguous .m files (Objective-C or MATLAB)
• Add reference and graph extraction for Ada

Evidence

languages reports gaps for Ada, Cython, D, Julia, MATLAB, Nim, and ambiguous_m (Objective-C or MATLAB).

Expected behavior / direction

Add bounded language-aware call/import/type/module relationships and fixtures. For .m, preserve the content-based disambiguation from the earlier fix and expose authoritative capabilities after classification rather than treating ambiguous input as structurally empty.

Related closed issues

• Dynamic language support should add symbols for Crystal, Groovy, Julia, and Tcl #3528 added Julia symbols.
• Systems language support should add symbols for Ada, D, and Nim #3529 added Ada/D/Nim symbols.
• Cython and CUDA files should expose native-extension and kernel symbols #3530 added Cython symbols.
• Disambiguate .m and .pl by content instead of assigning one language unconditionally #4612 added .m content disambiguation; it remains closed.

These issues remain closed; this is a new regression or follow-up observed on v1.39.4.

Environment

Reproduced with a locally built cdidx v1.39.4 from current origin/main at 87b8bf44210344f52ba4cb7827c149dfad32c0c8, using a fresh repository index on macOS arm64.

[Widthdom]

Work started on branch fix-issue4738.