[12.x] Remove calls to optional()

[cosmastech]

Saying goodbye to a relic from a bygone era.

We will always remember the good times we shared with optional(), the laughs and the tears. Listening to tunes on our iPods, drinking Four Lokos, and reaching for values which may or may not exist on a mixed type variable.

[bert-w]

Any plans to @deprecated the original optional() function? I myself have never used it and always preferred language constructs.

[shaedrich]

Well, after all, if people still want to use it, no harm in that

[jasonmccreary]

While I can't remember off the top of my head, years of Shift support lead me to believe there is a boundary case which optional covers that the nullsafe does not. Not speaking to this PR specifically, just generally commentary.

[rodrigopedra]

@jasonmccreary

I didn't use optional() for years.

Sometime ago, assessing an older codebase, I asked the developer why he kept using it, and he told me this:

• The underlying Optional class is macroable. So for lots of the DB façade calls without models, he could "add" methods to the \stdClass results:

  Optional::macro('toArray', fn() => (array) $this->value);
  
  \dump(\optional(\literal(name: 'Rodrigo'))->toArray());

• And, for some reason, he also often used the second $callback argument, as it only applies the callback when the value is not null

Might have to do with those cases.

I imagine it would be difficult to identify the first case in an automated code migration (calling a macro that might not have been defined in an obvious spot).