JIT doesn't trace exec()-generated functions called from hot loops

When a hot loop calls functions generated via exec(), the loop body gets traced but the callees don't. Each callee is individually cold (called ~60 times), but the call site is hot (5,000+ calls total across different callees).

Concrete case: tinygrad generates ~5,000 pattern matching functions at startup via exec(). They're called from a tight loop:

for _, match, _ in pats:
    if (ret := match(uop, ctx)) is not None: return ret

PYTHON_JIT=1 on 3.14.4 produces no improvement. The specializer works (LOAD_ATTR_SLOT fires correctly), but the JIT never compiles the callees.

Related: faster-cpython/ideas#738, #118093 (tier 2 entry at function calls).

It would be useful to either trace through exec()-generated callees at hot call sites, or detect aggregate hotness across callees sharing a call site.

Linked PRs

• gh-149575

Comece reproduzindo o loop de pattern matching do tinygrad com PYTHON_JIT=1 no Python 3.14.4, com foco nos callees gerados por exec() e em seu call site hot compartilhado. Compare o comportamento de LOAD_ATTR_SLOT do specializer com o fato de os callees serem compilados; considera-se concluído quando a carga de trabalho de chamadas hot relatada mostrar melhoria de JIT por meio das funções geradas.