Implement formulaparser.m

[AvanishSalunke]

though a previous initiative was taken to build this parser , #363

Here is a refined formulaparser.m that does support the fit models and anova function.

This is what has been implemented in short :

1. A lexer function (run_lexer) to convert the input strings into tokens.
2. A parser function (run_parser) to build the tree and support the precedence rules.
3. A expander function (run_expander) for creation of big eexplicit terms.
4. A schema builder function (run_schema_builder) which constructs a binary term matrix at the end.
5. A matrix builder (run_model_matrix_builder) to build the design matrix at the end.

Demos, documentation, BISTs have been added lately.

Not sure about the position of formulaparser in the INDEX file.

[pr0m1th3as]

I think parseWilkinsonNotation or parseWilkinsonFormula is a more appropriate name for this function. The statistics package already has a parseScoreTransform private function, so it makes sense to keep this kind of naming convention for specialized parser functions.

Can you add support for ^ as an "OP_POWER" operation and multiple return variables (repeated measures), as in "y1-y5 ~ x1:x2" or "y1,y4,y5 ~ x1^3"?

[AvanishSalunke]

@pr0m1th3as yes sure!! will come up with the changes.

[Pasta-coder]

you opened this PR without referencing the roadmap we established in Issue #266 or the ongoing discussion in #363. Since I am actively working on the LinearModel class structure which relies on this parser, it would have been better to coordinate to avoid overlapping efforts.
Also , this PR lacks removing partially implemented parser in GAMs (this was discussed in #266)

[pr0m1th3as]

you opened this PR without referencing the roadmap we established in Issue #266 or the ongoing discussion in #363. Since I am actively working on the LinearModel class structure which relies on this parser, it would have been better to coordinate to avoid overlapping efforts. Also , this PR lacks removing partially implemented parser in GAMs (this was discussed in #266)

The PR #363 is stalled because I explicitly asked you to stop pouring AI slop into that PR, but you kept going in the same direction. I haven't had the time to update the contribution guidelines yet, but let me be crystal clear about. I will not accept or tolerate any kind of contribution in any of the packages I maintain if I see any signs that it is the result of vibe coding.

[AvanishSalunke]

Hello @pr0m1th3as the changes have now been completed, firstly since "-" was already deletion operation and commas in LHS were not processed, now the formula is split into LHS and RHS and processed likewise. I also aded a new helper function to create list of the variables on the LHS side and the RHS side is processed as per the previous implementation but as ~RHS