Vector preprocessing happens during initialization rather than after calling .featurize() (Should it be moved?)

[xehu]

While testing out a local instance of the FeatureBuilder, I found that the preprocessing for vectors/sentiments happens upon initialization, rather than after calling .featurize(), which would make more sense.

Here's the output after merely initializing the FB:

>>> df_test = pd.DataFrame({
... "conversation_num": [1,2,3],
... "speaker_nickname": ["foo","foo","bar"],
... "message": ["lorem", "ipsum", "dolor"] 
... })
>>> my_fb = FeatureBuilder(input_df = df_test)
Initializing Featurization...
Confirmed that data has conversation_id_col column: conversation_num!
Confirmed that data has speaker_id_col column: speaker_nickname!
Confirmed that data has message_col column: message!
Generating RoBERTa sentiments...
100%|██████████████████████████████████████████████████████████████| 1/1 [00:00<00:00,  1.55it/s]

Was this the behavior all along, or did we previously start processing only after calling .featurize()? Creating this issue to document/address.

[xehu]

Update: seems like this has been the behavior since 0.1.1:

See this code blob: https://github.com/Watts-Lab/team_comm_tools/blob/b6c3767cff1ddbc5df93bac1806abd2da92fcd30/src/team_comm_tools/feature_builder.py

Looks like check_embeddings has always gotten called during init(). Converting this issue into a question (do we want to change this behavior? Is this annoying/unexpected for people?).