Sanitize _ in words/labels

[drkameleon]

Description

Basically, _ cannot be an identifier/word (because it simply isn't) - if it's followed by a space or a different symbol, etc... it has to be considered a standalone symbol.

Also see: #1773 (similar-ish)

Type of change

• Code cleanup
• Enhancement (implementation update, or general performance enhancements)

[Eloitor]

Could this be a step in the direction to get negative numbers like _1, _2 or _3?

[drkameleon]

Could this be a step in the direction to get negative numbers like _1, _2 or _3?

The logic is a bit different... and more for reasons of consistency: simply put, _ is not a "word". Now, considering _1 as "not a word" would be tempting too, but given that these are totally valid identifiers e.g. in JSON, this would start degrading the compatibility between what Arturo supports and different (and common) data formats. In face of the long-awaited (still proof of concept) Arturo-as-Data, we definitely don't want to completely spoil it.

(And one could say why not _ - which is also a valid JS identifier? Because, I guess we'd have to set some limit - even though it might seem arbitrary - and exclude things that go against the very core logic in Arturo: if it looks like a symbol, then it's most definitely not a "word" lol)

A couple of sidenotes:

• For negative numbers, as a literal, I'm still leaning in favor of your own unary-minus idea...
• The reason behind this PR - apart from a general effort of cleaning up ambiguously-parsed values - is this: https://discord.com/channels/765519132186640445/829324913097048065/1277266472120619171 (it's not going to be exactly like this, but our case supporting pattern matching could be more feasible and visually-ok than it initially looked 😉 )