* commit '80f934af':
  Add comKeyStyle

Bug: 10822107
Change-Id: I411a2427e759e83366bbf7a525b9ab8a1ab3144d

* commit '7ba78687':
  Add Ver4DictDecoder.

* commit '11ce0545':
  Fix hitbox of emoji key

Bug: 9618601
Change-Id: I43c5840505c6a847aaf4893a400392ccd45903c0

* commit '926ee64b':
  Add a new option for version 4 to dicttool.

* commit '4c5e6634':
  Groundwork for implementing GC.

Bug: 10780368
Change-Id: If10c18fdb0e7f0b0460a721d55001ab0fef28664

* commit 'ff493744':
  Quit skipping user history dictionary for gesture.

Bug: 6669677

Change-Id: Ia2b2f338795ab62cc9eef73731285b2a5a0ac5e4

* commit '6f68b736':
  Fix an off-by-one error.

* commit '284e73bd':
  Show emoji key on PC layout

These variables were meant to represent the (x,y) coordinate of the
last pixel of the grid, to avoid going one pixel too far. But the last
pixel of the grid, of course, has coord (width - 1, height - 1).

This bug only ever happens in a rare case because to reproduce,
it needs the right (resp. bottom) coordinate of the rightmost
(resp. bottommost) plus the threshold plus half the grid width
(resp. height) to be exactly the number of pixels of the grid,
which is pretty unlikely... unless there are more tiles in the
grid than pixels on the screen, in which case the grid width is
1 and this becomes likely.

This fixes the fencepost error and renames the variables to
clarify their meaning.

Bug: 10537485
Change-Id: Iecfe2c0f29ee17776ee2cceaa4b1db722276b1f3

* commit '142d68cf':
  Disable Nepali layouts temporarily per product call

Bug: 10812167
Change-Id: Ie1e22a1651f8ec40e4078c01d161d6c113de582e

Bug: 6669677
Change-Id: Id4344f2a30e9827f210055d6c65ee7776d3a63e3

bug: 6910787
bug: 9833890
Change-Id: Id58cdf117d145c03c96f8737a12bb1b11bd4e636

Change-Id: I18fd48c1f6921758d30330fbc77f4a917c33f1c8

* commit '7635ada2':
  Fix test in SubtypeLocaleUtilsTests

Change-Id: I4f7bf70ec7a0a67ad5d2129afed314b1b1a93a6c

* commit '86780628':
  Fix an NPE.

* commit '08bc97f8':
  Allow double-space-to-period after a plus char

Apparently, it is possible that File#getPath() returns null.
In this case, we'll have a null AssetFileAddress and an NPE
later down the road.

Bug: 9944824
Change-Id: Id6ceac1cb55439660f9811cf0fe3b574200d5a20

Bug: 10704936
Change-Id: I30fdaea648605ef7e7fea6c093581ed5c6de03f3

* commit '75feb0f6':
  Change name of header_reading_utlis.*

Bug: 6669677
Change-Id: I91389050148237580ef640d3a58b935ea752954e

* commit '561a8311':
  Add dictionary header writing methods.

Bug: 6669677
Change-Id: I392ac4776b55779903cbaa17e683005d80017a41