[FEATURE REQUEST] Customize domains for sources

[ElliotK-IB]

Similar issue:

• [feature request] Adding an option to source documents from Arxiv #96

Customizing allowed source domains would be helpful in cases when a user is interested in only material from specific sources. E.g.,

• *.gov, *.edu
• pubmed.ncbi.nlm.nih.gov/*
• arxiv.org/*

Note: Asking for only .gov or .edu in the prompt did not work for me.

I haven't looked deeply into this codebase but maybe something like "site:edu" could be injected into the search process? Or as was pointed out in #96, this filter step could be modified.

In addition to customizing a whitelist/allowlist, customizing the blacklist may also be useful.