Add Rio Trees to Simulator

[polybeandip]

This tweaks the simulator to be closer to Rio's semantics in math-land. Namely, it introduces

• riotree.ml, the Rio tree data structure
• Make_PIFOControl and Make_RioControl, functors that makes PIFO and Rio controls from Rio programs
• Make_Sim, a functor that takes a control and makes a function for running it on a PCAP

With these additions, tests/test_enq_deq_equiv.ml offers emperical evidence of the equivalence of our enqueue and dequeue side semantics! To generate Gantt charts from these tests,

1. run dune test with the environment variable GEN_CSV set to YES: i.e. GEN_CSV=YES dune test
2. run the script plot.py with python3 plot.py

Additionally, the simulator can also be used by running bin/sim.ml with

dune exec bin/sim.exe <prog> <pcap> <RIO|PIFO>

There are a few other minor contributions; here's a longer log of changes not mentioned above:

• make plot.py generate Gantt charts from all CSVs in graphs/; remove hardcoded names
• add programs to test enqueue-dequeue side equivalence in progs/enq-deq-equiv
• remove various type aliases in frontend/ast.ml
• make Policy.t look more like Ast.stream
• remove FIFO removals (i.e. extract_subpol) from Add Union Operator; Update AST #69
• tweak tests/frontend/test_parsing in accordance with changes to Policy.t
• convert Pieo and Pieotree modules back to Pifo and Pifotree
• delete mli files when everything is exposed anyway: e.g. rank.mli, time.mli, topo.mli
• expose certain types for ease of use: e.g. Rank.t, Time.t
• purge code for T2T compilation: large chunks of topo.ml
• close Policy.of_program does not always terminate #73 by fixing Policy.of_program
• tweak additions from Compiler: convert policy to JSON #75 to work with changes to Policy.t
• Make tests for both modules Riotree and Pifotree in test_rio_tree.ml and test_pifo_tree.ml
• Make exception LexerError and have ParserError hold a record of relevant info instead string

[polybeandip]

@anshumanmohan sorry for the large diff (again); a good deal of this is mostly minor refactoring though.

The footprint of the important changes (dequeue side semantics stuff) is almost entirely localized to

• riotree.ml(i)
• control.ml(i)
• simulate.ml(i)
• test_rio_tree.ml
• test_enq_deq_equiv.ml

Of the smaller tweaks, I had a couple deletions I hoped to run by you.

[polybeandip]

@anshumanmohan Just a heads up: I've merged @csziklai's work from #75 so this PR is ready for review again!

I've also marked some questions that came up for me during the merge

[anshumanmohan]

Thanks!!

[anshumanmohan]

Am I blocking this again, or is ready to go?

[polybeandip]

@anshumanmohan Finally hitting the big green button, sorry it so long!