enhance: optimistic search page loading and improved content refinement

[njhale]

Improves Google Search page loading, scraping, and LLM-based content refinement.

This PR includes the following enhancements:

• Wait for result page DOMs to stabilize -- i.e. no changes for 500ms -- before scraping content
• Do a better job pre-filtering non-content HTML elements before transforming to markdown
• Produce dense/prettified markdown content on the initial scrape
• Use tiktoken to get a more accurate token count for content truncation
• Change content refinement prompt to produce output suitable for truncation (dense markdown instead of JSON)
• Distribute the Google Search tool output token budget across results based on their size and the quality of their content when truncating refined search result content (the goal is to keep shorter, higher quality, content as intact as possible)

Requires the gptscript module in Obot to be bumped to a version that includes: #415 (see obot-platform/obot#1680 for bump PR)

Also addresses: obot-platform/obot#1671 and obot-platform/obot#1423

[g-linville]

Suggested change

	3. Compose a concise markdown document containing the excerpts organized in decending order of importance to understanding the topic. Do not paraphrase, summarize, or reword the excerpts. The goal is to preserve as much of the original content as possible.
	3. Compose a concise markdown document containing the excerpts organized in descending order of importance to understanding the topic. Do not paraphrase, summarize, or reword the excerpts. The goal is to preserve as much of the original content as possible.

[njhale]

Thanks for catching that! Pushed.