Unify and refactor NVDA in-process code

[jteh]

All of NVDA's in-process code should be unified into NVDAHelper. This will reduce duplicated code and greatly ease the process of adding new in-process code. NVDAHelper will handle the setting of system hooks and components will then register hook callbacks with NVDAHelper to be executed when the hook is fired. nvVBufLib will be merged into NVDAHelper, which means it will no longer be a separate library. This is becoming far too difficult to maintain and the advantages greatly outweigh the disadvantages.

Most of this work is already complete. However, we should be tracking these issues properly. :)

[jteh]

Bzr branch:  http://bzr.nvaccess.org/nvda/nvdaHelperRewrite/
This also fixes some crashes in nvVBufLib due to better detection of when a buffer should be terminated.

[jteh]

(In #381) Bzr branches:

• NVDA:  http://bzr.nvaccess.org/nvda/adobeAcrobatEnhancements/
• nvVBufLib:  http://bzr.nvaccess.org/nvVBufLib/adobeAcrobatEnhancements/

[aleksey_s]

(In #387) Fixed in bzr nvdaHelperRewrite r2932.
Btw, I am thinking about organizing our node-specific code somehow. May be separate code for each node into functions and put them into array or map...

[jteh]

Merged in r3110.