Add Ser/De protocol benchmarks

[RanVaknin]

Benchmarking notes

These benchmarks isolate the serialization/deserialization layer by attempting to use the lowest level public marshalling/unmarshalling layer in each SDK. Due to architectural differences between v1 and v2, the deserialization benchmarks have one minor asymmetry worth noting:

For JSON/RestJson/CBOR deserialization, v2's narrowest unmarshalling entry point takes an SdkHttpFullResponse, so the v2 benchmarks create an SdkHttpFullResponse wrapper per iteration. v1's narrowest entry point takes a raw parser, so it doesn't have this overhead. Even though v2's deserialization benchmark is slightly disadvantaged, it is faster across the board.

Throughput

Protocol	Operation	V1 (ops/µs)	V2 (ops/µs)	Diff
RestJson	CreateFunction deser	0.068	0.176	+159%
RestJson	CreateFunction ser	0.468	0.284	-39%
JSON	DynamoDB PutItem deser	0.072	0.141	+96%
JSON	DynamoDB PutItem ser	0.139	0.101	-27%
CBOR	CloudWatch GetMetricData deser	0.279	0.382	+37%
CBOR	CloudWatch GetMetricData ser	0.719	0.416	-42%
EC2	DescribeInstances deser	0.014	0.019	+36%
EC2	DescribeInstances ser	1.724	1.068	-38%
Query	STS AssumeRole deser	0.037	0.038	+3%
Query	STS AssumeRole ser	3.181	1.394	-56%
RestXml	CloudFront CreateDistribution deser	0.004	0.005	+25%
RestXml	CloudFront CreateDistribution ser	0.047	0.079	+68%

[sonarqubecloud]

Quality Gate failed

Failed conditions
0.5% Coverage on New Code (required ≥ 80%)
3.1% Duplication on New Code (required ≤ 3%)

See analysis details on SonarQube Cloud