Cleanup workshop notebooks

[antonum]

Summary of changes:

• Use HFTextVectorizer.embed_many(), instead of raw sentence_transformer consistently
• Replace Unstructured with PyPDF
• Cosmetic changes. The output is displayed as a df table instead of a raw object dump where possible.

[tylerhutcherson]

Looks like the CI failures here are because of the github actions secrets missing from Anton's fork. Looks like his changed notebooks are all good. Merging and will address any issues if they come up.