fix: shorten output filenames with readable labels and hashes

[kyuz0]

Long target URLs and dataset names can make generated filenames unreadable or too long to create. Compact options now use up to 24 readable characters plus a 12-character SHA-256 suffix, replacing punctuation such as =, /, and : with hyphens. Long dataset names and tags use the same helper; generated result and dataset filenames stay under 180 characters with their timestamps and extensions intact.

Automatic resume uses the same naming logic and also recognizes existing filenames. Full options still reach the target unchanged. The change is limited to output naming, resume lookup, documentation, and regression tests.

Validation:

• Ruff formatting and lint checks pass.
• All 401 functional tests verified: 398 passed in the sandbox; the three multiprocessing tests passed when rerun with local socket access.
• Functional tests cover long URLs, Unicode and punctuation, distinct hashes, filename length, old/new resume matching, and CLI dataset generation in all three formats.
• CLI coverage verifies that the target receives the original options and that the compact result file can be resumed.