Simplify get citation call for markup

[flooie]

Add Document object for better citation parsing

With the introduction of markup parsing,
handling multiple parameters became unwieldy.
This PR creates a new Document object that encapsulates:

• Plain and markup text
• Span updates for text mapping
• Tokenized words and extracted citation tokens

This should simplify function calls and ultimately enable better HTML extraction.
Also simplify the get citations when using markup text only requiring markup text and cleaning steps

Currently we must pre-process plaintext before parsing markup.

[grossir]

This looks great; just a couple small comments

[flooie]

Im going to suggest we just skip the benchmark because of the difficulty of changing the function calls.

[grossir]

We should change the benchmark call now, if not done, it will basically deactivate the benchmark for every PR until we actually fix it
Change isn't too long, it's on these lines

eyecite/benchmark/benchmark.py

Lines 55 to 67 in ba84b13

	if text:
	# Remove XML encodings from xml_harvard
	text = re.sub(r"^<\?xml.*?\?>", "", text, count=1)
	opinion_text_is_marked_up = True
	else:
	text = row["plain_text"]
	opinion_text_is_marked_up = False
	plain_text = clean_text(text, ["html", "inline_whitespace"])
	found_citations = get_citations(
	plain_text,
	markup_text=text if opinion_text_is_marked_up else "",
	)

            if text:
                # Remove XML encodings from xml_harvard
                markup_text = re.sub(r"^<\?xml.*?\?>", "", text, count=1)
                plain_text = ""
                clean_steps =  ["html", "inline_whitespace"]
            else:
                plain_text = row["plain_text"]
                clean_steps = ["all_whitespace"]
                markup_text = ""

            found_citations = get_citations(
                plain_text,
                markup_text=markup_text, 
                clean_steps =clean_steps
            )
            

[flooie]

im loving the performance gain- thanks I was working on the benchmark but looks all set now

[github-actions]

The Eyecite Report 👁️

Gains and Losses

There were 0 gains and 0 losses.

Click here to see details.

id	Gain	Loss

Time Chart

Generated Files

Branch 1 Output
Branch 2 Output
Full Output CSV

[mlissner]

How'd y'all squeeze out that performance???