MAINT Enable ruff PERF rule for performance linting

[romanlutz]

• Add PERF to ruff select list
• Ignore PERF203 (try-except in loop is intentional)
• Fix PERF401: use list comprehensions/extend instead of for-loop appends
• Fix PERF402: use list() instead of unnecessary comprehension
• Fix PERF403: use dict comprehension instead of for-loop updates
• Fix PERF102: use .keys()/.values() instead of iterating items()
• Add noqa for PERF401 in auxiliary_attacks (external code)

[Copilot]

Pull request overview

This maintenance PR enables the ruff PERF rule set (perflint) for performance-oriented linting and addresses all resulting violations across the codebase. The changes are purely mechanical refactors with no behavioral differences.

Changes:

• Adds PERF to the ruff select list in pyproject.toml (and ignores PERF203 for intentional try-except in loops)
• Fixes PERF401/402/403/102 violations: replaces for-loop-with-appends with list comprehensions, unnecessary comprehensions with list(), for-loop dict updates with dict.update(), and .items() iterations that only use values with .values()
• Adds a # noqa: PERF401 comment in external GCG attack code to suppress the rule there

Reviewed changes

Copilot reviewed 29 out of 29 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
pyproject.toml	Adds PERF to select and PERF203 to ignore
pyrit/auxiliary_attacks/gcg/attack/base/attack_manager.py	Converts for-loop to list comprehension + adds (misapplied) noqa comment
pyrit/analytics/conversation_analytics.py	Replaces manual loop+append with list comprehension
pyrit/executor/attack/printer/markdown_printer.py	Replaces multiple loop+append patterns with .extend() and generators
pyrit/executor/promptgen/fuzzer/fuzzer.py	Replaces manual loop+append with list comprehension
pyrit/identifiers/component_identifier.py	Replaces loop dict-update with dict comprehension + update()
pyrit/memory/memory_exporter.py	Replaces two manual loop+appends with list comprehensions
pyrit/memory/memory_interface.py	Replaces loop+append with conditions.extend()
pyrit/models/scenario_result.py	Replaces loop+append with objectives.extend()
pyrit/prompt_converter/audio_*.py (3 files)	Replaces loop+append for channels with list comprehension
pyrit/prompt_converter/nato_converter.py	Replaces loop+append with list comprehension
pyrit/prompt_target/openai/openai_chat_target.py	Replaces loop dict-update with dict.update()
pyrit/prompt_target/openai/openai_response_target.py	Replaces loop+append with list comprehension
pyrit/scenario/core/scenario_strategy.py	Replaces loop+append with .extend() and generator
pyrit/scenario/scenarios/airt/*.py (3 files)	Replaces loop+append with list comprehensions
pyrit/scenario/scenarios/foundry/red_team_agent.py	Replaces loop+append with list comprehension
pyrit/scenario/scenarios/garak/encoding.py	Replaces loop+append with list comprehension
pyrit/score/scorer.py	Replaces loop+append with list comprehension
pyrit/score/scorer_evaluation/scorer_metrics_io.py	Replaces loop dict-update with dict comprehension + update()
pyrit/setup/initializers/pyrit_initializer.py	Replaces .items() iteration (only using keys) with direct dict iteration
tests/unit/datasets/test_jailbreak_text.py	Replaces .items() where value was unused with .values()
tests/unit/executor/attack/test_attack_parameter_consistency.py	Replaces loop+append with .extend() and generator
tests/unit/registry/test_base_instance_registry.py	Replaces loop+append with list()
tests/unit/setup/test_load_default_datasets.py	Replaces loop+append with .extend() and generator
tests/unit/target/test_playwright_target.py	Replaces loop+append with list comprehension

[Copilot]

The # noqa: PERF401 comment on this line is misapplied. PERF401 flags for-loops that build lists using .append() and can be replaced with a list comprehension. However, line 110 IS already a list comprehension, so PERF401 would never fire on it in the first place.

The PR description states the intent was to "Add noqa for PERF401 in auxiliary_attacks (external code)". The correct approach for preserving external code as-is would be to keep the original for-loop and add the # noqa: PERF401 there. Instead, the code was converted to a list comprehension (changing the external code) AND the now-meaningless noqa was added.

The noqa comment should either be removed (since the list comprehension doesn't trigger PERF401) or the original for-loop form should be restored with the noqa suppression, consistent with the stated intent of not modifying external code.

Suggested change

	ascii_toks = [i for i in range(3, tokenizer.vocab_size) if not is_ascii(tokenizer.decode([i]))] # noqa: PERF401
	ascii_toks = [i for i in range(3, tokenizer.vocab_size) if not is_ascii(tokenizer.decode([i]))]

[Copilot]

The PERF rule prefix was added to select but not to fixable. Looking at other multi-character rule codes in the fixable list (PIE, RET, SIM, TCH, etc.), there's a consistent pattern of including selected multi-character rule codes in fixable as well. Without PERF in the fixable list, ruff --fix will not auto-fix any future PERF violations, requiring manual fixes every time. Consider adding "PERF" to the fixable list for consistency.