Optimize memory usage for TSDataset._merge_exog

[brsnw250]

Before submitting (must do checklist)

• Did you read the contribution guide?
• Did you update the docs? We use Numpy format for all the methods and classes.
• Did you write any new necessary tests?
• Did you update the CHANGELOG?

Performance graphs (before)

Time

Memory

Proposed Changes

• Updated exog variables addition to the dataset.
• Reworked reggressors checks
• Memory saved: 6.8G -> 5.1G (1.7G -- 25%)
• Memory consumption is more uniform accross the timeline
• Speed up: 3:59 -> 0:22 (~11x)

Performance graphs (after)

Time

Memory

Code to reproduce

from etna.datasets import TSDataset, generate_ar_df

df = generate_ar_df(n_segments=100_000, periods=100, start_time="2000-01-01")

df_exog = generate_ar_df(n_segments=100_000, periods=110, start_time="2000-01-01")
df_exog.rename(columns={"target": "exog_0"}, inplace=True)
df_exog["cap"] = 101

for i in range(1, 10):
    df_exog[f"exog_{i}"] = df_exog["exog_0"].values + i

print("Creating dataset")
ts = TSDataset(df=df, df_exog=df_exog, known_future="all", freq="D")
print("Done.")

Closing issues

closes #403

[github-actions]

🚀 Deployed on https://deploy-preview-596--etna-docs.netlify.app

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 90.42%. Comparing base (7508561) to head (36cb957).
Report is 1 commits behind head on master.

Additional details and impacted files

@@           Coverage Diff           @@
##           master     #596   +/-   ##
=======================================
  Coverage   90.41%   90.42%           
=======================================
  Files         259      259           
  Lines       18022    18036   +14     
=======================================
+ Hits        16295    16309   +14     
  Misses       1727     1727           

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

[d-a-bunin]

Optimized -> Optimize