Raise TypeError when a dictionary is passed to MeasurementResult constructor

[natestemen]

Motivation: Multiple students at the UMass Amherst summer school attempted to pass a dictionary to instantiate a MeasurementResult object while working through the challenges. This silently accepts, but provides erroneous results. E.g.:

>>> from mitiq import MeasurementResult
>>> MeasurementResult({'001': 23, '000': 43})
MeasurementResult: {'nqubits': 3, 'qubit_indices': (0, 1, 2), 'shots': 2, 'counts': {'001': 1, '000': 1}}

Considerations: I could see the following two potential solutions to help with this issue.

1. Add a typecheck at runtime and raise a TypeError, letting the user know we have a method MeasurementResult.from_counts for this specific pattern.
2. Call MeasurementResult.from_counts in the __post_init__ method to re-instantiate self.

I chose to raise solution 1 as it seems to be a simple approach to address the problem. Admittedly, I am less familiar with the technical details of re-instantiating an object within the __post_init__ function, but happy to test that out if reviewers think it is more appropriate.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 98.72%. Comparing base (411e234) to head (b6d77e9).
Report is 5 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #2523   +/-   ##
=======================================
  Coverage   98.72%   98.72%           
=======================================
  Files          90       90           
  Lines        4156     4158    +2     
=======================================
+ Hits         4103     4105    +2     
  Misses         53       53           

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

[cosenal]

Solution 1 sounds reasonable to me, and a strict improvement on the current behavior.

Note: in the docstring, we should have an example of how to instantiate MeasurementResult, at the moment it's not very clear (unless I am overlooking it).

[cosenal]

We could have something stricter that checks for result not being an instance of Sequence[Bitstring], which is what we expect as type, but then both the check and the error message become more complicated, so I would leave it as is.

[natestemen]

Thanks for the suggestion of adding an example in the docstring. I like that. Done in 71a0a63.