Summary of the issue:
When activating Highlight of navigator object, the highlighting rectangle does not show up. It only shows up when the navigator object is moved for the first time. If the highlight of nav object is previously disabled, the issue occurs for all the four following types of activation:

When checking the corresponding checkbox in the vision parameters
When you reload NVDA's saved config provided that highlighter is enabled in NVDA's saved config.
When activating with a gesture a profile containing nav object highlight enabled
When NVDA starts up if the default config has nav object highlight enabled
Description of how this pull request fixes the issue:
Checks nav object initialization when the providers are initialized. Prior to this PR, only the focus initialization was checked (and the caret thanks to it).

…, allow panning after review commands (#13718)

Fixes #8682
Fixes #11519

Summary of the issue:
#8682: when using the read_currentLine, read_currentWord, and read_currentCharacter commands after the caret has been moved, it is impossible to pan the Braille display through the displayed content.
#11519: when moving the mouse with Mouse Tracking on (default) and Braille Tether set to Auto (default), using read_currentLine, read_currentWord, and read_currentCharacter speaks content under the mouse, but does not update the Braille display with the spoken content.

Description of how this pull request fixes the issue:
These three global review functions all call braille.handler.setTether early in their processing. braille.handler.setTether clears the display buffer, making it impossible to scroll through the displayed content. braille.handler.setTether also does not update the display, causing the problems discussed in #11519 -- tether flags are set properly, but the display does not change. A simple call to braille.handler.handleReviewMove in all three cases solves both of these issues.

… position information (#13744)

Fixes #9184 and fixes #9728 

Summary of the issue:
A number of features in Java Access Bridge were not working optimally.
1. Various commands like read window and read window title were not working when Java applications have multiple windows. (#9184)
2. Toggle buttons were not read correctly. (#9728)
3. Unnecessary spaces when HTML tags removed from Java controls. (#9728)
4. Provide position information for Java tab controls. (#9728)

Description of how this pull request fixes the issue:
1. According to Java documentation* when getAccessibleIndexInParent returns -1 there is no parent. So in NVDA if indexInParent is None we should treat it as if parent is not set.
2. Java toggle buttons use the checked state for pressed, correctly map this. The value should be ignored for a toggle button.
3. Updated the regular expression processing for HTML tags to check if the tags are at the start or end or have surrounding whitespace, in which case the tag is stripped. If there is only non-whitespace characters around the tag then replace the tag with a space to prevent words being joined.
4. For Java controls with Role.TAB use indexInParent and parent.childCount to find the position information.

* https://docs.oracle.com/en/java/javase/17/docs/api/java.desktop/javax/accessibility/AccessibleContext.html#getAccessibleIndexInParent()

…tire code base. (#13646)

Summary of the issue:
Process ID of NVDA is retrieved and used inconsistently in the code base:

In appModuleHandler we're storing it as a module level variable when initializing, however it is also pointlessly reinitialized when user re-initializes appModules - this makes no sense as process ID is constant during NVDA's life time.
Most modules retrieve it by calling os.getpid
In some cases win32 function GetCurrentProcessId is used directly.


Description of how this pull request fixes the issue:
NVDA process ID is stored in globalVars when NVDA starts - all usages of os.getpid and GetCurrentProcessId now use the new variable. Since appModuleHandler.NVDAProcessID is gone appropriate backward compatibility solution for add-ons has been implemented.

### Summary of the issue:
Liblouis 3.22.0 has been released.

### Description of how this pull request fixes the issue:
Updates liblouis to 3.22.0. Changelog: https://github.com/liblouis/liblouis/releases/tag/v3.22.0

### Testing strategy:
Ran from sources and loaded the new braille table. Works as it should.

…t required (#13761)

Fixes #13659, #13735

Summary of the issue:
Certain applications don't provide a column order array. For the case of single column list views, a column order array is unnecessary, as a default mapping can be assumed.

Description of how this pull request fixes the issue:
For single column list views, map the first column in the UX, to the first column index: i.e. [0] is the column order array. When fetching from a multi column list view, log an error if the column order array is unknown.

Closes #13727

Summary of the issue:
eSpeak fails to perform automatic language switching for a common language codes "en", "fr",
Reported to eSpeak project: espeak-ng/espeak-ng#1200
A PR has been opened against eSpeak to fix this: espeak-ng/espeak-ng#1211

Description of how this pull request fixes the issue:
Creates a dictionary of aliases for language codes without locale, which map to appropriate supported languages with locale.
This may need to be added to over time as appropriate.
Integration tests have been added to ensure this list is comprehensive when we update eSpeak.
When eSpeak is unaware of a language code, the new behaviour is to:
• check for a default, if not then
• check for the language stripped of the dialect locale, if not then
• check for any matching language (regardless of dialect, e.g 'en-au' to 'en-us'), if not then
• ignore the language change

Fixes #13732

Summary of the issue:
OneCore keeps a cache of voices for NVDA in the registry.
OneCore fails to automatic language switch when remnants of a previously installed OneCore language exist in the cache.
This causes speech to not be announced.

This is only an issue for installed copies, as portable copies bypass the NVDA OneCore cache.

Description of how this pull request fixes the issue:
Drop the language change command if a given language is unsupported by OneCore.