/usepmp command no longer accepts profile IDs with non-alphanumeric characters (including the default UUIDs generated when creating a persona)

[Kagamiin]

Describe the bug

After a recent update that changed the behavior of the /usepmp command, it is no longer accepting profile IDs that contain characters other than A-Z, a-z, 0-9 and _ (underscore) - which includes the automatically-generated UUIDs when creating a persona through the settings menu (as they contain dashes), as well as certain mnemonics for members of plural systems (such as mine which also happens to contain a dash).

Reproduction

1. Ensure per-message-profile support is enabled in the Experimental tab of the settings menu.
2. Create a persona in the Personas tab in the settings menu - it'll automatically be assigned a UUID for its ID; you can either leave it as-is, or choose a different string of characters that happens to contain a non-whitespace non-alphanumeric character that isn't the underscore (_). Give it a name, save it and copy the ID assigned to it.
3. Attempt to run the /usepmp command in a room, pasting in the ID that was assigned to the new persona as an argument.
4. Having executed the command, attempt to send a message under the newly-assigned persona.

Expected behavior

The command should execute properly for any set of non-whitespace characters you can input as a persona ID in the Personas tab in the settings menu (at the very least but not limited to any printable non-whitespace ASCII characters), and the persona should be selected as the current PMP according to the specified criteria.

Platform and versions

1. OS: Arch Linux (KDE/Wayland/nvidia-580xx, linux-zen 6.19.6)
2. Browser: libcef (Sable 0.0.1/Electron)
3. Sable version: v1.13.0 (709a9a9)
4. Matrix homeserver: chat.blahaj.zone

Additional context

This seems to be caused by the regular expression that's being used to parse the command's payload, at 155ac64#diff-0e4172772c62327a8689d866202697fb752960019446deaee231bcedf5bc3079R55:

const USEPMP_REGEX = /^(\w+)\s*(-g)?(-o)?(-u)?\s*(\d+)?$/;

The \w+ token being used to parse the command payload is equivalent to [A-Za-z0-9_]+ and will make the regular expression not match anything that contains characters other than the aforementioned ones.

I consider this to be a regression as prior to this change, there were no issues running /usepmp with at least the default UUIDs generated by the Persona tab in settings, and at least one user was able to successfully use a colon in a persona ID without any issues.

[dozro]

yeah there's an error in the regex, need to fix that. thanks for reporting :)

the switch to regex instead of how it was before that, is partly a small step towards #156