Fix CMR validation when pyQuARC is used as a library

[slesaad]

Problem

The CMR validation feature added in v1.3.0 (#364) breaks every concept-id validation when pyQuARC is imported as a package — which is how QuARC's Lambda uses it. The QuARC staging API currently returns HTTP 500 for all concept-id requests because of this.

Two separate failures:

1. _validate_with_cmr() references CONTENT_TYPE_MAP, but that constant is only imported in the if __name__ == "__main__" branch of main.py. Library usage raises:

   NameError: name 'CONTENT_TYPE_MAP' is not defined
   

2. With that fixed, cmr_response.json() raises JSONDecodeError for every collection that passes CMR validation, because CMR's ingest validate endpoint returns HTTP 200 with an empty body on success (it only returns a JSON body when there are errors).

Fix

• Import CONTENT_TYPE_MAP in the package-import branch.
• Parse the CMR response defensively: a non-JSON/empty body is treated as no errors/warnings. This also drops the duplicated .json() call.

Verification

Ran in a python3.9 Amazon Linux 2 container (matching the QuARC Lambda runtime) against live CMR:

arc = ARC(input_concept_ids=["C1214470488-ASF"], metadata_format="echo-c",
          cmr_host="https://cmr.earthdata.nasa.gov")
results = arc.validate()
# concept_id: C1214470488-ASF, 85 checked fields,
# cmr_validation: {"errors": [], "warnings": []}, pyquarc_errors: []

Before the fix, the same call raises the NameError above. Full test suite: 27 passed.

🤖 Generated with Claude Code