FIX: Reject a blob URL without a container instead of raising IndexError

[Chen Yufeiyang (feiiiiii5)]

Description

AzureBlobStorageIO.parse_blob_url documents ValueError: If file_path is not a valid blob URL, but only the missing-scheme / missing-host case raises it. A URL with a scheme and host and no path (https://account.blob.core.windows.net) reached parsed_url.path.split("/")[1] and raised IndexError: list index out of range, and a URL ending in / produced an empty container name instead.

_resolve_blob_name catches ValueError only, so its fallback to treating the input as a local path never ran for either input, and every entry point that goes through it (read_file_async, write_file_async, path_exists_async, delete_file_async, is_file_async) inherited the IndexError.

The path is now checked the same way the scheme and host are: a container segment that is missing or empty raises ValueError("Invalid blob URL").

Tests and Documentation

Test: pytest tests/unit/memory/storage/test_storage.py -q -k "without_container_path or empty_container_name or without_container_falls_back" fails on main — IndexError at storage.py:309 for two of them and DID NOT RAISE for the third — and passes here. pytest tests/unit/memory/storage -q → 128 passed; the wider set (storage + test_azure_sql_memory + the Azure Blob target) → 235 passed, 1 skipped. ruff check and ruff format --check clean; ty check reports the same single pre-existing diagnostic at storage.py:97 before and after.

No documentation change beyond what the docstring already promised.