Add a configuration option to parse HTML based on CSS selectors

[JohnGUnderwood]

When using the 'web' data load the chunking will sometime split text in the middle of a sentence. I'd like to have an option to pass in a css selector that pulls out a list of paragraphs from the source page. Then each complete paragraph could be its own chunk.

See an example here: https://github.com/JohnGUnderwood/atlas-news-search/blob/main/backend/packages/main.py#L177-L198

In the screenshots below the source article content is an array of paragraphs extracted from HTML using selectors. Each item is then turned into its own document with embedding.