Fix the compiler suggesting unimplemented variables

[Daniel-Cortez]

What this PR does / why we need it:

This PR fixes the problem where after the 1'st pass the compiler suggests global variables before the point of their definition:

GetValue()
{
    return value; // error 017: undefined symbol "value"; did you mean "value"?
}

new value = 0;

main()
{
    GetValue();
}

Which issue(s) this PR fixes:

What kind of pull this is:

• A Bug Fix
• A New Feature
• Some repository meta (documentation, etc)
• Other

Additional Documentation:

[YashasSamaga]

not related specifically to this PR but to the system

There is an assert(0) in error_suggest but doing tagof(X) where X is a string, number, etc. leads to that assertion because of error_suggest(20,st,NULL,estNONSYMBOL,tok);. Since tok is passed as subtype, it can be tRATIONAL, tNUMBER, tSTRING, etc. depending on what the user entered.

The previous version of the compiler was giving out errors instead of an assertion failure.

[Y-Less]

I have wished you could take the tag of literals a few times:

tagof (X:5)

Not obviously useful as typed, but is in macro outputs. But I have macros to remove anything after the : for now.

[Daniel-Cortez]

Ok, I fixed this and a couple of other bugs locally. Would it be alright if I turn this PR into a more general one and upload those bugfixes here? (Just asking because this PR has already been approved in the current state.)

[Y-Less]

No, make those a separate pull request. The point of approvals is to show that the code's been checked, adding more after that defeats the purpose.

[Daniel-Cortez]

Well, I wasn't asking this only because this PR has already been approved, but also because uploading those fixes in a separate PR would definitely lead to a mess with merge conflicts, but ok.

[YashasSamaga]

I think every new feature should have a corresponding issue. I couldn't find an issue for this system which is why I sneaked it in here. I could have made an issue instead of commenting it here though.