test: Use valid enum values in graphqlapi_api_cache test

[licjun]

Issue #, if available

N/A

Description of changes

bin/run_cfn_lint.sh runs cfn-lint -u before linting, which pulls the latest CloudFormation schema. That schema now restricts AWS::AppSync::ApiCache ApiCachingBehavior and Type to enum values, so the placeholder values in the graphqlapi_api_cache test fail lint with E3030 on every PR, regardless of what the PR changes:

tests/translator/output/graphqlapi_api_cache.json:36:9:36:29:E3030:'somebehaviour' is not one of ['FULL_REQUEST_CACHING', 'OPERATION_LEVEL_CACHING', 'PER_RESOLVER_CACHING']
tests/translator/output/graphqlapi_api_cache.json:46:9:46:15:E3030:'mytype' is not one of ['LARGE', ..., 'SMALL', ...]

(plus the same two errors in the aws-cn and aws-us-gov outputs).

SAM passes both properties through unchanged (passthrough_value in sam_resources.py, PassThroughProp in the schema source), so the test only needs some value to verify the passthrough. This replaces the placeholders with valid values (FULL_REQUEST_CACHING, SMALL) in the input and the three partition outputs. No samtranslator code changes.

Description of how you validated changes

• pytest tests/translator/test_translator.py -k graphqlapi_api_cache: 3 passed (all partitions).
• bin/run_cfn_lint.sh (cfn-lint 1.55.0, schema updated in the run): exit 0, no errors. Before this change it reported the 6 E3030 errors above.

Checklist

• Review the generative AI contribution guidelines
• Adheres to the development guidelines
• Add/update transform tests
  • Using correct values
  • Using wrong values
• Add/update integration tests

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.