MeasurementResult Predicates

[johnzl-777]

Since #308 measurement now returns a MeasurementResult type instead of defaulting to a boolean type.

While SquinToStim handles this without a problem (it just checks the condition has a proper MeasureIdBool from the measure_id analysis) it does make type inference problematic because you run into cases where one could write:

measurement_result : MeasurementResult = squin.qubit.measure(q[0])
if measurement_result: # this is not a boolean! 
	squin.z(q[0])

To remedy this there have to be predicate statements (probably added to the qubit dialect) that can consume a MeasurementResult and query specific information about it to give a bool that can be properly inferred with. (Credit to @cduck who I believe brought this up in an older meeting).

For example, the above problematic code would be fixed via:

measurement_result: MeasurementResult = squin.qubit.measure(q[0])
if is_one(measurement_result):
	squin.z(q[0])

# which can be interpreted as CZ rec[<insert negative int here>] 0

The IfToStim rewrite rule would have to be slightly tweaked so that it only rewrites scf.IfElse if the condition's parent is is_one but that should just be a few additional lines of code.

I believe it would be best for #549 to be addressed first before proceeding with any of this.

[johnzl-777]

I think it's probably better for #580 to be completed first and then this can be brought in on top. The new type lattice/modified analysis will affect some of the operation of the physical/annotate + squin to stim pass