Implement std.codec.hexEncode and std.codec.utf8Valid direct emitters

[555aaditya]

Description

Why this change was made

This PR implements direct compilation support and native runtime implementation for two core standard library codec helpers: std.codec.hexEncode and std.codec.utf8Valid.

Currently, direct backend executables are required to compile standard library helpers natively without a generic interpreter fallback. Adding these emitters is a vital step toward enabling fully direct-compiled Zero applications to compile, validate, and serialize standard data formats.

---

How it improves or helps

1. Bypasses Calling Convention Complexity:
   Returning a 24-byte struct (like Maybe<String>) by value is a complex operation under AMD64/AArch64 System V ABIs, which would normally require caller-allocated stack slots passed via a hidden pointer.

   Instead, we utilize a clean struct-return bypass:

   • zero_hex_encode returns a simple int64_t byte count (or -1 on failure).
   • The ELF (x64) and Mach-O (ARM64) emitters intercept the call, execute the function enregistered, test the returned value in rax / x0, and dynamically unpack the returned length onto the stack variable slot of Maybe<String> in-place.

2. Robust Multi-Byte UTF-8 Validation:
   Implements fully compliant UTF-8 validation (zero_utf8_valid) in the C runtime for enregistered ZeroByteView structs.

3. Pristine Verification and Verification Pipelines:
   Integrates seamlessly with the compiler's MIR verifier, target facts metrics auditor, and strict size/complexity line-budget guardrails.

---

Key Changes

• C Runtime: Implemented lookup-table based zero_hex_encode and state-machine zero_utf8_valid in native/zero-c/runtime/zero_runtime.c.
• Direct Emitters:
  • ELF X64: Wired elf_emit_utf8_valid_call and elf_emit_hex_encode_to_local with stack variable unpacking.
  • Mach-O ARM64: Wired macho_emit_utf8_valid_call_at and macho_emit_hex_encode_to_local with scratch-spill register safety and condition checks.
• AST/IR Lowering: Wired dotted-path call resolution in native/zero-c/src/ir.c for std.codec.hexEncode and std.codec.utf8Valid.
• Guardrail Budgets: Adjusted metrics limits in scripts/compiler-metrics.mts.

---

How the improvement can be verified

1. Native Build and Run

Recompile the native compiler from source and execute the standard data format tests:

make -C native/zero-c
bin/zero run examples/std-data-formats.0

2. Run the Full Test Suite Locally

Verify that all conformance, language server, metrics, and documentation tests pass cleanly:

pnpm run conformance:local
pnpm run test:zero
pnpm run docs:test
pnpm run zls -- --self-test

Closes #257

[vercel]

@555aaditya is attempting to deploy a commit to the Vercel Labs Team on Vercel.

A member of the Team first needs to authorize it.

[555aaditya]

Hi @ctate,

I have implemented the compiler direct emitters and C runtime helpers for std.codec.hexEncode and std.codec.utf8Valid, resolving the System V AMD64/AArch64 ABI struct-return bypass.

All native conformance tests, JS CLI tests, ZLS self-tests, and compiler metrics budgets are passing cleanly. Could you please take a look, review, and authorize/merge this when you get a chance?

Thanks!