Use line as a unit for which say all is moved in rich edit controls

[lukaszgo1]

Link to issue number:

Fixes #13420
Fixes #9179

Summary of the issue:

When in say all NVDA moves caret by the unit specified as a RREADINGCHUNK for a given text info implementation. In most text info implementations it defaults to line, however in ITextDocumentTextInfo it was set to sentence. While it probably results in smoother reading for literary text written in language where TOM can split into sentences properly, in text's such as log or for languages without clear concept of sentence such as Chinese it results in caret not following spoken text since it is moved to the current position only when the sentence ending is reached.

Description of user facing changes

Say all both for review and system cursor moves by line in Rich Edit controls such as WordPad or NVDA's log viewer.

Description of development approach

for ITextDocumentTextInfo textInfos.UNIT_READINGCHUNK has been set to line.

Testing strategy:

Made sure that say all both with system caret and review cursor follows spoken text in QRead and log viewer.

Known issues with pull request:

• I'm not sure why the textInfos.UNIT_READINGCHUNK was set to the sentence in the first place, perhaps @michaelDCurran or @jcsteh would be able to recall the reason
• Ideally we would be able to move caret in say all with a higher precision e.g. by word/character - that is tracked in Say all, move caret per word instead of per line/sentence, taken over after abandoned 9937 #11658

Change log entries:

Bug fixes

• Caret properly follows spoken text in say all in places such as bookworm, WordPad, or NVDA's log viewer (NVDA can't make system cursor and review cursor follow by using say all in WX's rich text box, Take NVDA log viewer and bookworm as examples #13420 When performing say all in the log viewer cared isn't following spoken text. #9179)

Code Review Checklist:

• Pull Request description:
  • description is up to date
  • change log entries
• Testing:
  • Unit tests
  • System (end to end) tests
  • Manual testing
• API is compatible with existing add-ons.
• Documentation:
  • User Documentation
  • Developer / Technical Documentation
  • Context sensitive help for GUI changes
• UX of all users considered:
  • Speech
  • Braille
  • Low Vision
  • Different web browsers
  • Localization in other languages / culture than English

[jcsteh]

I don't recall this specific case, but usually, when we used sentence, it was either because the underlying implementation didn't support line at the time or because line caused a performance problem. If it works now, it seems reasonable to change it for consistency if nothing else.