Improve learn_rounding_digits precision

[fealho]

CU-86b455jpu, Resolve https://github.com/datacebo/SDV-Enterprise/issues/1107.

A float64 has 15-17 digits of precision. Anything after that should be disconsidered. For example:

f"{np.float64(0.1):.15f}" -> '0.100000000000000'
f"{np.float64(0.1):.20f}" -> '0.10000000000000000555'

These 15-17 digits include the integer part of the number, for example:

f"{np.float64(100.1):.12f}" -> '100.100000000000'
f"{np.float64(100.1):.15f}" -> '100.099999999999994'

The previous implementation always attempted to round to MAX_DECIMALS - 1 digits, ie 14 decimal places, which could fail for cases like above where there's an integer component to the number:

np.float64(944343.1221).round(13) -> 944343.1220999999
np.float64(944343.1221).round(12) -> 944343.1221

This PR fixes this issue by rounding to MAX_DECIMALS - num_integer_digits instead, ensuring the situation above never happens.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 100.00%. Comparing base (e7bacd2) to head (b103696).
Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff            @@
##              main      #964   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files           18        18           
  Lines         2352      2353    +1     
=========================================
+ Hits          2352      2353    +1     

Flag	Coverage Δ
integration	83.08% <100.00%> (+<0.01%)	⬆️
unit	100.00% <100.00%> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[sdv-team]

Task linked: CU-86b455jpu SDV-Enterprise - HSA Synthesizer fails to properly learn rounding precision for some numerical columns #1107

[R-Palazzo]

Have you tried the changes on the colab notebook of the issue?

[R-Palazzo]

LGTM!
Great test to understand and double check everything 👍

[fealho]

Have you tried the changes on the colab notebook of the issue?

Added a notebook showing it fixes the original problem here.