Faster duplicate check

[MarshalX]

No description provided.

[codspeed-hq]

CodSpeed Performance Report

Merging #34 will degrade performances by 27.64%

_{Comparing faster-duplicate-check (ebe67c4) with main (f19b9bc)}

Summary

⚡ 6 improvements
❌ 4 regressions
✅ 182 untouched benchmarks

⚠️ Please fix the performance issues or acknowledge them on CodSpeed.

Benchmarks breakdown

	Benchmark	main	faster-duplicate-check	Change
⚡	test_dag_cbor_decode[roundtrip11.json]	34.2 µs	25.1 µs	+36.23%
❌	test_dag_cbor_decode[roundtrip25.json]	27.2 µs	34.9 µs	-22.12%
⚡	test_dag_cbor_decode_fixtures[dagpb_simple_forms_2]	98.4 µs	85.2 µs	+15.45%
❌	test_dag_cbor_encode[roundtrip04.json]	24.4 µs	33.7 µs	-27.64%
⚡	test_dag_cbor_encode[roundtrip08.json]	33.7 µs	24.7 µs	+36.57%
❌	test_dag_cbor_encode[roundtrip09.json]	38.7 µs	46.9 µs	-17.47%
❌	test_dag_cbor_encode[roundtrip13.json]	24.6 µs	33.8 µs	-27.2%
⚡	test_dag_cbor_encode[roundtrip17.json]	27 µs	24.5 µs	+10.05%
⚡	test_dag_cbor_encode[roundtrip24.json]	33.4 µs	24.3 µs	+37.82%
⚡	test_dag_cbor_encode[roundtrip26.json]	33.5 µs	25.2 µs	+33.13%

[MarshalX]

it doesnt give noticeable perf improve because we don't have key duplication in valid data :) so get_item always (in good cases) return None which is as fast as contains