feat(badger): cx and rz const functions and strategies for LexicographicCostFunction

[lmondada]

A couple of points to note:

1. I've made some minor breaking changes to the Rust API of LexicographicCostFunction. I think it is cleaner now.
2. I had the choice between keeping fn pointers as the cost function type within LexicographicCostFunction or moving to Box<Fn>. I've stuck to the former for the moment, but I didn't figure out a simple way to reuse the same code for Tk2Op::CX and Tk2Op::RzF64 without using closures. The current code has some duplication as a result, but I think it's bearable.
3. I've tried running badger with cost_fn='rz', but the Rz gate count does not decrease at all. I've looked for an obvious bug but I don't think it is within these changes...

Let me know if you disagree with 1. or 2 and what you think we should do about 3.

---

Changelog metadata

BEGIN_COMMIT_OVERRIDE
feat: BadgerOptimiser.load_precompiled, BadgerOptimiser.compile_eccs and passes.badger_pass now take an optional cost_fn parameter to specify the cost function to minimise. Supported values are 'cx' (default behaviour) and 'rz'.

END_COMMIT_OVERRIDE

[lmondada]

Here's some code that shows no optimisation is happening:

from tket2.passes import badger_pass
from pytket import Circuit, OpType

import json

with open("test_files/barenco_tof_10.json", "r") as f:
    circ = Circuit.from_dict(json.load(f))

p = badger_pass(cost_fn="rz", log_dir="badger_logs", timeout=10)

print('Rz count before:', circ.n_gates_of_type(OpType.Rz))
p.apply(circ)
print('Rz count after:', circ.n_gates_of_type(OpType.Rz))

output:

Rz count before: 100
Rz count after: 100

:(

EDIT: This might just be because barenco_tof_10 has already been optimised for T count!

[codecov]

Codecov Report

Attention: Patch coverage is 37.50000% with 25 lines in your changes missing coverage. Please review.

Project coverage is 81.69%. Comparing base (295b0df) to head (304d45c).
Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
tket2/src/rewrite/strategy.rs	42.30%	15 Missing ⚠️
tket2/src/optimiser/badger.rs	28.57%	10 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #625      +/-   ##
==========================================
- Coverage   82.06%   81.69%   -0.37%     
==========================================
  Files          48       48              
  Lines        6572     6606      +34     
  Branches     6572     6606      +34     
==========================================
+ Hits         5393     5397       +4     
- Misses        820      850      +30     
  Partials      359      359              

Flag	Coverage Δ
rust	81.69% <37.50%> (-0.37%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[lmondada]

Update: Running the above script on a different input does something, for example using qf21_n15 and timeout=10 results in a Rz count decrease from 166 to 149.

(Make sure to rebase to CX + H + Rz first)

However using cost_fn='cx' performs equally well, so still not convinced it is doing the right thing.

Running FullPeepholeOptimise followed by Badger (timeout=30) reduces Rz gate 166 -> 122 -> 118.

[aborgna-q]

I edited the changes to make them non-breaking (deprecated the renamed method instead)