Cutting decompose step should return a subexperiment dictionary

[caleb-johnson]

For circuit cutting, users should have a way to map a subexperiment index (output from the decompose step) to the specific sample/sub-observable from which it was created. The sample could be represented in a number of ways -- maybe a list of (basis, id) tuples would be an intuitive representation of a "sample"?

[garrison]

This is a continuation of the conversation at #378 (comment):

I think [that ordering the observables like sample_{0}observable_{0}, sample_{0}observable_{1}, ..., sample_{0}observable_{N}, ..., sample_{M}observable_{N}] is fine as a first step, but I would really like to move away from this and instead expose an explicit map that allows looking up each (sample, observable) from the index in this flat list and vice versa. The reason is because some experiments are actually redundant (#262) or potentially even unnecessary, i.e. if the only observables in that subregion are all Pauli Is (in which case those experiments can be skipped altogether).

[caleb-johnson]

This issue is stale. Dictionaries are now used in cutting-for-width workflows.