docs: clarify get/first/last MultipleMatches semantics

[collerek]

Summary

Fixes the internally inconsistent docstrings flagged in #967.

• first() / last() (and their _or_none variants) always run with LIMIT 1 via _fetch_single, so MultipleMatches can never be raised — the docstrings claimed otherwise.
• get() enforces the single-match contract only when criteria are set (via filter/exclude or args/kwargs). With no criteria it falls back to returning the last row by pk (LIMIT 1) and cannot raise MultipleMatches either. The old wording ("first row meeting the criteria … MultipleMatches if more than 1 row") conflated these two modes.

Docs-only change: updated docstrings in ormar/queryset/queryset.py and ormar/relations/querysetproxy.py, plus the user-facing docs/queries/read.md sections for get / first / last.

Closes #967

Test plan

• make pre-commit
• make coverage (100%, 653 passed)

[codspeed-hq]

Merging this PR will not alter performance

⚡ 3 improved benchmarks
❌ 1 (👁 1) regressed benchmark
✅ 94 untouched benchmarks

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
⚡	WallTime	test_get_or_create_when_create[40]	483.4 ms	417.2 ms	+15.89%
⚡	WallTime	test_update_or_create_when_create[10]	83.7 ms	75.2 ms	+11.26%
👁	WallTime	test_saving_models_individually[20]	133.3 ms	160.6 ms	-16.98%
⚡	WallTime	test_updating_models_individually[40]	325.8 ms	285.5 ms	+14.14%

Tip

Curious why this is faster? Comment @codspeedbot explain why this is faster on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing docs/clarify-get-first-multiple-matches (fc1ddb1) with master (260c6a7)¹}

Footnotes

1. No successful run was found on master (722595f) during the generation of this report, so 260c6a7 was used instead as the comparison base. There might be some changes unrelated to this pull request in this report. ↩