refactor: Remove metakey concept

[Xuanwo]

Which issue does this PR close?

Part of #5314

Rationale for this change

To make Lister behavior more predictable.

What changes are included in this PR?

Remove the metakey concept.

Are there any user-facing changes?

[codspeed-hq]

CodSpeed Performance Report

Merging #5319 will degrade performances by 16.59%

_{Comparing remove-metakey (11ef737) with main (297bbf9)}

Summary

⚡ 4 improvements
❌ 10 regressions
✅ 59 untouched benchmarks

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

Benchmarks breakdown

	Benchmark	main	remove-metakey	Change
❌	bytes buf 256 KiB * 32 advance	435 ns	493.3 ns	-11.82%
❌	bytes buf 256 KiB * 4 advance	435 ns	493.3 ns	-11.82%
❌	bytes buf 4.00 MiB * 32 advance	435 ns	493.3 ns	-11.82%
❌	bytes buf 4.00 MiB * 4 advance	435 ns	493.3 ns	-11.82%
⚡	256 KiB * 1000k chunk	183.6 ns	154.4 ns	+18.88%
❌	256 KiB * 1000k truncate	245.8 ns	275 ns	-10.61%
⚡	256 KiB * 100k chunk	183.6 ns	154.4 ns	+18.88%
❌	256 KiB * 100k truncate	245.8 ns	275 ns	-10.61%
⚡	256 KiB * 10k chunk	212.8 ns	183.6 ns	+15.89%
❌	256 KiB * 10k truncate	245.8 ns	275 ns	-10.61%
⚡	256 KiB * 1k chunk	183.6 ns	154.4 ns	+18.88%
❌	256 KiB * 1k truncate	245.8 ns	275 ns	-10.61%
❌	concurrent 16	299.8 µs	359.5 µs	-16.59%
❌	concurrent 8	319.2 µs	363.9 µs	-12.3%

[Xuanwo]

Merging #5319 will degrade performances by 15.89%

cc @dqhl76, I'm considering disabling tests for those cases or simply refactoring them. They've never been reported correctly...

Any ideas?

[dqhl76]

Any ideas?

Feel the same with you. It can be a bit distracting. For those who aren't familiar with what it means, it might cause unnecessary concern.

[tisonkun]

LGTM