feat: Parallelizing test generation phase

[dhl123]

Summary

smith --flag test_generation runs four LLM stages — decompose → grey_condition → variable_extraction → case_generation. With BATCH_PROCESSING=true, each split its input into batches and sent one gateway request per batch in a serial loop, waiting for each response before issuing the next.

Now they are parallelized exactly as translation does. This applies the pattern established in extract_tool_args.py to the generation side.

• Output order is preserved.
• GENERATION_CONCURRENCY=1 (or -1) reproduces the old serial path exactly — no pool is created
• Concurrency above the batch count is harmless — max_workers is a ceiling, and threads are created lazily

Closes: #84

Changes

File	Change
src/smith/test_generation/concurrency.py	New — resolve_generation_concurrency() + shared run_batches()
src/smith/test_generation/decompose.py	Serial loop → run_batches; intra-batch pairing moved inside the worker
src/smith/test_generation/grey_condition.py	Serial loop → run_batches
src/smith/test_generation/variable_extraction.py	Serial loop → run_batches
src/smith/test_generation/case_generation.py	Serial loop → run_batches
src/smith/cli.py	Resolve GENERATION_CONCURRENCY once, thread it through generate_test
tests/integration/fakes.py	FakeOpenAI thread safety + content-keyed by_prompt mode
tests/integration/test_generation_unit.py	+9 tests
.env_template	GENERATION_CONCURRENCY=4
docs/content/docs/configuration.md	New row; fixed the TRANSLATION_CONCURRENCY default inconsistency
CLAUDE.md	Configuration-model prose
CHANGELOG.md	modified ### Added entry

Checks

• make ci passes (lint, Rego lint, license headers, build smoke)
• make test passes (policy scorecard — needed if policy behavior changed)
• CHANGELOG.md updated under ## [Unreleased] (if user-facing)
• PR title uses a conventional prefix (feat:, fix:, test:, docs:, chore:, or refactor:)
• Commits are signed off for the DCO (git commit -s)

Notes (optional)

Design sketch, screenshots, or extra context.

[araujof]

@dhl123 can you please merge main and resolve conflicts?

[araujof]

LGTM