Implement: multiple save and load for mlflow registry

[yleilawang]

1. Implemented save_multiple and load_multiple for mlflow registry
2. Test cases for implementation.

[codecov]

Codecov Report

Attention: Patch coverage is 59.37500% with 13 lines in your changes missing coverage. Please review.

Project coverage is 92.02%. Comparing base (b18b2d2) to head (69caf21).

Files with missing lines	Patch %	Lines
numalogic/registry/mlflow_registry.py	59.37%	10 Missing and 3 partials ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #416      +/-   ##
==========================================
- Coverage   92.20%   92.02%   -0.18%     
==========================================
  Files          98       98              
  Lines        4834     4865      +31     
  Branches      437      442       +5     
==========================================
+ Hits         4457     4477      +20     
- Misses        277      285       +8     
- Partials      100      103       +3     

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

[s0nicboOm]

Also, update the version for pyproject.toml

[ab93]

Also, update the version for pyproject.toml

@s0nicboOm we have not released any previous versions right?

[yleilawang]

Added changes for:

1. Changed return type of load_multiple to ArtifactData, the same return type of load.
2. Slots for CompositeModels class.
3. Handling exceptions caused by unwarpping pyfunc models.
4. Private function for getting unique sorted dkey list.
5. Added test cases for cache loading for pyfunc models.

[yleilawang]

Profiling Updates

reg.save(skeys=skeys, dkeys= ["ae", "pipeline"], artifact=VanillaAE(10), artifact_type='pytorch', **{"a": "b"})
reg.save(skeys=skeys, dkeys= ["scaler", "pipeline"], artifact=StandardScaler(), artifact_type='sklearn', **{"a": "b"})
reg.save(skeys=skeys, dkeys= ["threshold", "pipeline"], artifact=StdDevThreshold(), artifact_type='sklearn', **{"a": "b"})

output = reg.save_multiple(
    skeys=skeys, dkeys=dkeys, dict_artifacts=dict_artifacts, **{"a": "b"}
)

def f():
    reg.load(skeys=skeys, dkeys= ["ae", "pipeline"], artifact_type='pytorch')
    reg.load(skeys=skeys, dkeys= ["scaler", "pipeline"], artifact_type='sklearn')
    reg.load(skeys=skeys, dkeys= ["threshold", "pipeline"], artifact_type='sklearn')

def g():
    reg.load_multiple(skeys=skeys, dkeys=dkeys)

import timeit

a = timeit.timeit(f, number=100)
b = timeit.timeit(g, number=100)

print(a, b)

--> 9.074686333002319 4.829918750001525

[ab93]

Looks good!

[s0nicboOm]

Minor comments.
Otherwise LGTM!