Add rpcv2Cbor protocol test for multibyte UTF-8 text, integer-encoded floating point, and negative-integer timestamps - #3253

Background

Add rpcv2Cbor protocol test vectors that pin the on-the-wire CBOR encoding for previously untested cases: multibyte UTF-8 text (a string value and a map key, asserting the text-string length is a UTF-8 byte count, not a UTF-16 code-unit count); integer-encoded floating point (float/double deserialized from CBOR integers whose 2-, 4-, and 8-byte argument markers collide with the FLOAT16/32/64 minor values, asserting dispatch on the major type first); and negative-integer timestamps (pre-epoch values as CBOR negative integers, single- and multi-byte, asserting the sign is preserved).

These encodings are valid per RFC 8949 and the rpcv2Cbor spec but were not covered by the shared suite. The gaps let non-interoperable behavior pass unnoticed — the underlying defects survived round-trip tests because the same implementation encoded and decoded them consistently, so only wire-format assertions catch them. Pinning the wire bytes in the shared suite protects every SDK that consumes it against regressions.

Testing

• ./gradlew :smithy-protocol-tests:build — the model validates and the projection builds; each test body was decoded back to CBOR and checked against its params
• End-to-end against smithy-kotlin

Links

• Smithy-kotlin fix: Fix cbor serde smithy-kotlin#1677

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.