Add shots.bins() generator method

[Tarun-Kumar07]

Context:
The current approach for sampling multiple shots involves handling shot vectors by sampling multiple times and processing each shot independently. However, a more efficient strategy would be to sample all shots at once and then process each bin separately. The shots.bins() method facilitates this by providing the lower and upper bounds for each bin.

Description of the Change:

• The shots.bins() method generates a sequence of tuples, where each tuple represents the lower and upper bounds of a bin.
• In _measure_with_samples_diagonalizing_gates sample once with total number of shots and use shots.bins() to process each bin separately

Benefits:
This method is particularly useful for processing shot quantities in a partitioned manner, as it allows for the separate handling of each bin's range

Possible Drawbacks:

Related GitHub Issues:
Fixes #5433

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 99.66%. Comparing base (a47d9bc) to head (a9209ae).
Report is 4 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #5476      +/-   ##
==========================================
- Coverage   99.67%   99.66%   -0.01%     
==========================================
  Files         406      406              
  Lines       37881    37637     -244     
==========================================
- Hits        37758    37512     -246     
- Misses        123      125       +2     

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

[albi3ro]

For the PRNG key failures:

FAILED tests/devices/qubit/test_sampling.py::TestBroadcastingPRNG::test_nonsample_measure_shot_vector[measurement1-expected1-shots3] - assert False
 +  where False = <function allclose at 0x7f8312da2e70>(Array([-1.    ,  1.    , -0.0238], dtype=float64), array([-1,  1,  0]), atol=0.01)
 +    where <function allclose at 0x7f8312da2e70> = np.allclose
FAILED tests/devices/qubit/test_sampling.py::TestBroadcastingPRNG::test_nonsample_measure_shot_vector[measurement1-expected1-shots4] - assert False
 +  where False = <function allclose at 0x7f8312da2e70>(Array([-1.    ,  1.    , -0.0163], dtype=float64), array([-1,  1,  0]), atol=0.01)
 +    where <function allclose at 0x7f8312da2e70> = np.allclose
= 2 failed, 1066 passed, 108 skipped, 15 xfailed, 88 warnings in 737.38s (0:12:17) =

I think we will just have to change the test a little to accept the new results. Basically, we should expect the samples with shot vectors to be slightly different now, as sampling 2 then sampling 1 will give different values then sampling 3 at the same time, especially when PRNG keys are involved.

[albi3ro]

Looks great! Happy to finally have that alternate way of calculating shot vectors implemented.

My one comment was purely stylistic.

[albi3ro]

Requested a second review from the team.

[dwierichs]

Looks good to me, thanks @Tarun-Kumar07 ! 🎉

I had two small comments, one of which we briefly should discuss with @albi3ro .

[dwierichs]

Thanks @Tarun-Kumar07 💯