WIP: upgrade to SymbolicUtils w/ fast terms

[YingboMa]

This PR also deprecates .op and .args on Terms.

Most of the functionalities are upgraded except Hessian sparsity. @shashi is looking into that.

[ChrisRackauckas]

Is this breaking? @YingboMa take a look at Pumas and make sure it stays fine.

[isaacsas]

We'll need some Catalyst updates too.

[YingboMa]

Yeah, it's a little breaking, unfortunately. I will test those when this PR and the SymbolicUtils PR are finalized.

[ChrisRackauckas]

Yeah so this probably qualifies for a breaking release.

[ChrisRackauckas]

Yeah no worries. I am fine with having it a bit common in this stage in the repo, especially for a 100x performance gain. It just needs to be done right and take care of our downstream buddies

[YingboMa]

If we are going to do a breaking release anyway. Let's also sneak in #684 :P

[YingboMa]

All tests pass locally for me now.

[shashi]

@isaacsas @ChrisRackauckas (ping anyone else using this package downstream)

This is a breaking change! Let me summarize what this means to ModelingToolkit users:

• Term is not the only expression type now, there is also Add, Mul, and Pow.
• If you were dispatching on Term, now use istree(x) to check if x is of any of these expression types. All of them support operation and arguments accessors, so:
• x.op and x.args are deprecated in favor of operation(x) and arguments(x).
  Use these sed commands to make make this change across all files:
  • sed -i 's/\(\b[a-zA-Z0-9_]*\b\)\.args/arguments\(\1\)/g' **/*.jl
  • sed -i 's/\(\b[a-zA-Z0-9_]*\b\)\.op/operation\(\1\)/g' **/**/*.jl
    Note that expr.args where expr is an Expr will also be rewritten by this change, so be careful of what you commit (use git add -p to select the right changes, commit, and then run git reset --hard -- or however you usually do this.).

Here is what changed (and a migration guide) in SymbolicUtils.jl JuliaSymbolics/SymbolicUtils.jl#154 (comment) if you need to get into the finer details.

[ChrisRackauckas]

Thanks, makes sense. Let's make sure own downstream libraries all get PRs too.

[YingboMa]

I’d like to do some more breaking changes before we merge it, like changing the BipartiteGraph struct and the derivative.

[ChrisRackauckas]

Sounds good.

[YingboMa]

It's the annoying string LaTeX test again... Can we just delete it to test for nonempty string? Or only looking for some patterns.

[ChrisRackauckas]

what's going on there?

[YingboMa]

Oh, this is actually due to the exp change. I will fix it.

[YingboMa]

Fingers crossed. CI should pass in this commit.