abi: new v2: typed numeric storage

[edsiper]

This PR introduces typed scalar value support in cmetrics to preserve integer precision end-to-end (especially OTLP NumberDataPoint.as_int) while remaining backward-compatible with existing MsgPack payloads.

What changed

• Added typed scalar model to cmt_metric:
  • value type tag: double | int64 | uint64
  • exact integer storage fields
  • kept legacy packed-double field for compatibility paths
• Added typed scalar APIs:
  • cmt_metric_set_double, cmt_metric_set_int64, cmt_metric_set_uint64
  • cmt_metric_get_value_type, cmt_metric_get_int64_value, cmt_metric_get_uint64_value
  • existing cmt_metric_set/get_value still work (compat wrapper behavior preserved)
• Updated OTLP decode/encode:
  • decode as_int into typed int64
  • encode emits exact integer when representable, preserving roundtrip fidelity
• Updated MsgPack encode/decode ABI:
  • still writes/reads legacy "value" (old payload compatibility)
  • adds typed fields for new payloads: "value_type" + "value_int64"/"value_uint64"
  • decoder accepts old/new payloads and prefers typed fields when present
• Updated scalar copy/cat paths to preserve typed value semantics.

Tests

• Extended OTLP roundtrip coverage for large integer precision (> 2^53) across:
  • OTLP -> cmetrics -> MsgPack -> cmetrics -> OTLP
• Added standalone MsgPack ABI/security tests:
  • legacy value-only decode path
  • new typed-int64 decode path
  • malformed summary/histogram payload mismatch rejection tests
• Full suite passing:
  • ctest --test-dir build_test_otlp --output-on-failure (18/18)

Compatibility

• Backward compatibility with older MsgPack payloads is preserved.
• Existing API users relying on double setters/getters continue to work unchanged.
• Precision-sensitive integer pipelines now roundtrip losslessly where protocol allows.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: d1b6e73735

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".