Support appending/replacing text (parts) to a TextDisplay

[dhardy]

Summary

Adds lower-level methods for adding text to a TextDisplay, supporting adding a text in parts. Also supports replacing text parts.

Motivation

The existing design is a poor fit for text editors and potentially also for viewing large documents for performance reasons: it requires (re)-preparing all text at once.

This new API supports splitting a source text at line-breaks and adding each part independently, appending to or replacing old content.

Currently text-wrapping and alignment must still happen for all content afterwards; this could be addressed in the future (if necessary) but is relatively cheap.

[dhardy]

The motivation for this didn't quite work out for text editors since highlighting information also needs to be updated piecemeal, then matched up to the TextDisplay at run-time: it therefore makes sense to use a separate TextDisplay for each "piece" of text.

One (minor) feature is needed now: the ability to turn off imply_empty_final_line.

I will merge this PR anyway because (in my opinion) it fits well into kas-text and may eventually find some other uses.

API change summary

New pub methods on TextDisplay:

• clear(self)
• runs_len(self) -> usize
• append_runs(self) -> RunAppender
• replace_runs(self, range) -> RunReplacer`

Changed pub methods on TextDisplay:

• resize_runs has an additional range parameter; pass .. to achieve old behaviour

New pub struct RunAppender has pub methods:

• push_text(self, text, font_tokens, imply_empty_final_Line) -> Result<(), NoFontMatch>
• push_text_range(self, text, range, font_tokens, imply_empty_final_Line) -> Result<(), NoFontMatch>

New pub struct RunReplacer has pub methods:

• index() -> usize
• push_text(text, font_tokens, imply_empty_final_Line) -> Result<(), NoFontMatch>
• push_text_range(self, text, range, font_tokens, imply_empty_final_Line) -> Result<(), NoFontMatch>

New pub struct AnalyzedText has pub method:

• new(text, direction) -> Self