Use stable IDs for saved query filenames

[JohanLorenzo]

Changing a query title left duplicate slugged files, and cleanup sometimes removed the wrong copy (#61). I switched saved-query filenames to use the stable Redash ID. Fetch migrates legacy files only when their tracked content matches the server and preserves divergent files; a successful deploy writes the resulting server copy under the ID-only name.

Fixes #61.

Evidence

• stmo-cli fetch migrated a legacy {id}-{slug}.* pair to {id}.*. The migrated files matched the Redash copy byte-for-byte, and the legacy files were removed.
• stmo-cli fetch on a legacy pair with a local SQL edit exited 1 with "Local SQL or metadata differs from Redash". It left the legacy files unchanged and created no {id}.* files.

Fetching 1 specific queries...

[...]

✓ Fetched 0/1 queries
Error: Failed to fetch 1 query operation(s): 127140: Local SQL or metadata differs from Redash for query 127140; leaving queries/127140-old-slug.* unchanged. Resolve or deploy the local changes before fetching
exit=1