codegen: safety fallback + entry-slice boundary fix for truncated functions (#86)

[jlsandri]

Addresses #86. Two small, independent fixes for functions whose last instruction is not jr \$ra — either because the source really lacks a tail return, or because the imported function boundary is wrong and the real jr \$ra lives past the reported end.

Fix 1 — Implicit safety fallback at function end (code_generator.cpp)

Emit an implicit

ctx->pc = GPR_U32(ctx, 31);
return;

at the end of every generated function body.

• For well-formed functions this is unreachable dead code (the preceding jr \$ra already returned).
• For functions with incorrect boundaries, it returns cleanly via \$ra instead of leaving ctx->pc at whatever value the last fall-through instruction set, which previously caused cascading wrong returns up the call chain.

Fix 2 — Skip boundary starts inside a containing parent (ps2_recompiler.cpp)

When reslicing entry functions, skip any boundary start that falls inside the range of the containing parent function.

Ghidra-style sub_xxx children inside a parent were creating "boundary" entries that truncated sibling entry slices before their jr \$ra, producing half-functions that would fall off the end at runtime.

Scope

• ps2xRecomp/src/lib/code_generator.cpp: +6 / -0
• ps2xRecomp/src/lib/ps2_recompiler.cpp: +13 / -1
• No change in behaviour for correct inputs.

[jlsandri]

Closing as part of a batch cleanup after #107 landed. The runtime ecosystem refactor in #107 substantially reworked the files this PR touched, and I would like to re-audit the underlying fix against the new code structure before putting it back in front of you. If the fix is still needed after that re-audit, I will re-open as a focused PR rebased onto current main. Thanks for your patience.