fix: allow numeric resource IDs in _VALID_RESOURCE_NAME_REGEX

[urmzd]

Description

_VALID_RESOURCE_NAME_REGEX in the RAG SDK requires the first character to be a lowercase letter ([a-z]), which rejects bare numeric IDs (e.g. 1234567890) that the Vertex AI API assigns to resources like RAG corpora and files.

Calling get_corpus() or get_file() with a valid numeric resource ID raises ValueError:

from vertexai.preview import rag

corpus = rag.get_corpus("1234567890")
# ValueError: name must be of the format
#   `projects/{project}/locations/{location}/ragCorpora/{rag_corpus}` or `{rag_corpus}`

The ID is valid — parse_rag_corpus_path returns {} (not a full path), and the regex rejects it because "1" doesn't match [a-z]. The fix branch never reaches rag_corpus_path() to expand the short ID.

Fixes #6442

Changes

Updated the regex from [a-z][a-zA-Z0-9._-]{0,127} to [a-zA-Z0-9][a-zA-Z0-9._-]{0,127} in both definitions:

• vertexai/rag/utils/_gapic_utils.py
• vertexai/preview/rag/utils/_gapic_utils.py

Testing

• Added test_get_corpus_numeric_id_success and test_get_file_numeric_id_success to both test_rag_data.py and test_rag_data_preview.py
• Tests use isolated mocks with real side_effect on path helpers so bare numeric IDs exercise the regex code path (the shared fixture's Mock() always returns truthy for parse_rag_corpus_path, bypassing the regex entirely)
• All 201 RAG tests pass (4 new + 197 existing)