[kocaemre]

Summary

• Use the same medium HTTP timeout for iterate_keys() as list_keys() in both sync and async clients. This applies to every paginated key request; callers can still pass an explicit timeout.
• Add multi-page regressions for both clients and both supported HTTP transports.

Fixes #1173.

Verification

• RED: uv run --extra httpx2 pytest tests/unit/test_client_pagination.py -k 'kvs_iterate_keys_uses_list_keys_default_timeout' -q -o addopts='' --tb=short -> 4 expected failures (the page calls used long).
• GREEN: same command -> 4 passed; uv run --extra httpx2 pytest tests/unit/test_client_pagination.py -q -o addopts='' --tb=short -> 756 passed.
• uv run --extra httpx2 --extra brotli poe unit-tests -> 2412 passed, 1 skipped (upstream generator incompatibility on Python 3.15).
• uv run --extra httpx2 poe check-docstrings, uv run --extra httpx2 ty check src/apify_client/_resource_clients/key_value_store.py tests/unit/test_client_pagination.py, uv run --extra httpx2 ruff check src/apify_client/_resource_clients/key_value_store.py tests/unit/test_client_pagination.py, uv run --extra httpx2 ruff format --check src/apify_client/_resource_clients/key_value_store.py tests/unit/test_client_pagination.py, git diff --check -> passed.
• No live API integration tests; the regressions use a local pagination HTTP server.

AI assistance was used for implementation and tests; the change and verification were reviewed locally.