Investigate using tket to optimize for IonQ devices

[bachase]

To complement the findings in #461, this issue is to investigate cross-compiling via pytket with an appropriately defined IonQ backend https://docs.quantinuum.com/tket/extensions/pytket-quantinuum/index.html#default-compilation (https://github.com/CQCL/pytket-ionq is now archived). Given tket is built by Quantinuum, we want to see if it has better out-of-the box performance. This comparison would be added as another compiler in the notebook introduced in unitaryfoundation/research#32.

[bachase]

The (now deprecated) pytket-ionq provider does have an implementation that targets IonQ QIS gates, but not native gates.

To add support for IonQ native gates,

1. For any native gate not in Tket set, define a new custom gate
2. Create a RebaseCustom that decomposes the tket internal gateset of TK1,CX to these gates
3. Likely add a SquashCustom pass that can identify ways to combine the native gates

We can then still use this in a sequence pass like those defined here.

To validate these were applied correctly, we can validate against predicates like these

[bachase]

I'm marking this as nice to have, as the progress in #491 and the pending tket2 migration make this less a priority IMHO