Cardano Updates

- Consolidate API_DOCKER_IMAGE_TAG, INDEXER_DOCKER_IMAGE_TAG into RELEASE_VERSION
- Align mainnet paths with preprod pattern
- Add configurable container config path (#660)
- Update documentation

- Add version field to pyproject.toml for uv PEP 621 compliance
- Update test filename references from test_construction_api.py to test_golden_examples.py
- Mark protocol_parameters.protocol as volatile for cross-network compatibility

Recent uv versions enforce PEP 621 compliance requiring the version field.
Also renamed test_construction_api.py to test_golden_examples.py in workflows and README.

- Rename test_construction_api.py to test_golden_examples.py
- Add `endpoint` field support for explicit endpoint declaration
- Add `skip_reason` field for controlled test skipping
- Add /search/transactions ordering test case (#534)
- Document skipped tests with ticket references (#582, #585)

Move test_tokens_expose_policy_id from test_token_registry.py to test_account_endpoints.py as it validates API contract (policyId always present for native assets) rather than token registry enrichment.

The test was incorrectly passing when token registry was misconfigured because policyId is populated from blockchain data, not registry metadata.

Changes:
- Add TestNativeAssetMetadata class to test_account_endpoints.py
- Remove TestPolicyIdMetadata class from test_token_registry.py
- Enhanced validation to check all native assets, not just configured tokens
- Added policyId format validation (56 hex chars)

- Add data_type field to DIMENSIONS in populate_data.py
- Replace ADDRESSES/BLOCKS/TRANSACTIONS with ENDPOINT_DATA dict
- Add tx_history support to /search/transactions (address-based search)

- Make CARDANO_NETWORK env var required (no silent fallback to preprod)
- Refactor RosettaClient to use explicit network_identifier parameter
- Use typed parameters for all API methods
- Consolidate network_identifier error tests in test_error_handling.py
- Standardize all tests to use network_identifier dict pattern

- Remove hardcoded DIMENSIONS dict (70 lines)
- Load from dimensions.json in main() before any operations
- Change data_subdir → data_type to match dimensions.json schema

Replaces old profile-based testing with dimension-isolated approach to
identify specific performance bottlenecks in the Rosetta API.

New testing infrastructure:
- populate_data.py: Generate test data from Yaci Store DB organized by
  dimensions (utxo_count, tx_history, block_tx_count, etc.) with
  power-of-10 or percentile levels stored in dimensions.json
- stability_test.py: Capacity testing using Apache Bench with exponential
  search and dimension isolation. Supports dict-based CSV loading for
  flexible data formats (addresses, blocks, transactions)
- soak_test.py: Long-running stability tests using Locust
- analyze_dimensions.py: Statistical analysis of dimension distributions

Updated locustfile.py to load all dimension/level combinations from
dimensions.json instead of hardcoded light/medium/heavy categories,
enabling per-dimension performance tracking in Locust UI.

Key improvements from dimension isolation:
- Identifies which specific dimension causes degradation (UTXO count vs
  token count vs tx history)
- CSV format adapts to dimension type (addresses have 'address' column,
  transactions have 'transaction_hash,block_hash,block_index')
- Markdown reports group results by endpoint and dimension
- Transaction queries include block_index for /block/transaction testing

Deleted obsolete files: test_data.py, populate_test_data.py, main.py

- Fix script path from test_construction_api.py to test_golden_examples.py
- Remove exit 0 that masked golden example test failures
- Add explicit failure if API never starts in wait loop
- Always post PR comment, even when deployment fails
- Check all test suites (smoke, behavioral, golden) before reporting pass/fail