codegen: handle continuation PCs and nested callee yields

[jlsandri]

Stacked on #110 (codegen: remove dead else branch). Please merge #110 first — once it lands, this PR's diff shrinks to just the two commits below.

Problem

The recompiler's "continuation PC" model assumed a 1:1 mapping between continuation targets and emitted entry functions, and that calls into functions with continuation PCs would always yield back at the single expected fallthrough address.

Two related failure modes:

1. Continuation targets without a separate entry function. When the analysis produced a continuation PC that did not have its own entry_... wrapper, the codegen path assumed the entry existed and produced a call that couldn't be dispatched.
2. Nested callee yields. When callee A yielded internally and re-entered itself at a continuation PC, and callee B was called from inside A, B's own continuation yield would bubble up as a PC_MISMATCH in A's caller — a false positive, because A was still mid-execution and just hadn't reached its outer fallthrough yet.

Fix

Two commits:

1. Handle continuation PCs without separate entry functions. Teach code_generator.cpp / ps2_recompiler.cpp to dispatch continuation targets directly when no entry wrapper exists, instead of assuming the wrapper is always present. Adds a new bookkeeping field to the codegen header.

2. Add a continuation re-entry loop for nested callee yields. After a call, emit a while loop that keeps re-entering the callee while its yielded ctx->pc lies inside the callee's own range, only exiting to the caller when the real fallthrough PC is reached.

These go together — the re-entry loop depends on the first commit's bookkeeping to know where a continuation target lives.

Scope

• ps2xRecomp/include/ps2recomp/code_generator.h: +3 / -0
• ps2xRecomp/src/lib/code_generator.cpp: +124 / -6
• ps2xRecomp/src/lib/ps2_recompiler.cpp: +41 / -17

Rationale

Continuation PCs were already a supported concept in the recompiler; this PR just makes the handling robust to the two edge cases above, which were producing false PC_MISMATCH aborts in otherwise well-formed code.

[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.