fix(tinytorch): tito package reset is unreachable

[Shashank-Tripathi-07]

Bug

tito package reset (with or without piped input) does nothing but print a generic help panel every time, regardless of input. The documented example in PackageCommand's own help text, tito package reset --force, silently does nothing at all.

Root cause

ResetCommand was written as if it would be registered as its own top-level tito reset <subcommand> command: its name property returns "reset", and its own help text says tito reset progress --backup. It is actually only ever mounted at tito package reset. By the time argparse reaches this class's own add_subparsers() call, both the package and reset tokens are already consumed by the parent parser, so a bare tito package reset never has a SUBCOMMAND token left to match, and args.reset_command is always unset. The only way to actually trigger a reset was the undocumented double-noun tito package reset package --force.

Fix

• Added a --force argument directly on the reset parser itself (not just on each sub-subparser), so it's available for a bare invocation.
• Bare tito package reset [--force] (no SUBCOMMAND) now defaults to resetting the package, matching PackageCommand's own documented example and this command's stated top-level purpose ("Reset tinytorch package to clean state"). It still shows the other reset types (all/progress/milestones/config) for discoverability, then proceeds with the actual package reset (still gated behind its own confirmation prompt unless --force is passed).
• tito package reset package --force (the old double-noun form) still works unchanged.

Testing

Verified against a fresh dev install with all 20 modules exported:

• tito package reset, answering "n": shows the confirmation prompt and correctly cancels. Before the fix, this printed only the self-referential help panel and never reached a confirmation at all.
• tito package reset --force: correctly removes all 20 generated package files, matching the documented example exactly.
• tito package reset package --force: still works as before.

Test plan

• tito package reset, confirm it shows a real confirmation prompt (not just a help panel)
• tito package reset --force, confirm it actually removes exported package files
• tito package reset package --force, confirm the old form still works

[profvjreddi]

Thanks @Shashank-Tripathi-07. Integrated into dev, CI green.

One repair on the way in. Each subparser already defines its own --force, and argparse writes a subparser's default over a shared parent dest, so tito package reset --force package silently dropped the flag and then blocked on input(). Under a script or in CI that is an EOFError, which is the opposite of what someone passing --force wants:

$ tito package reset --force package < /dev/null
EOFError: EOF when reading a line

Gave the parent flag its own dest and OR the two in run(), so --force works on either side of the SUBCOMMAND token. run() also has to set args.force itself, since the bare form parses no subparser and the attribute otherwise doesn't exist.