Improve process output watcher's handling of multi-byte UTF8 characters

Speciffically, handle case when a non ASCII UTF8 character is read
partially from the process output. Instead of reporting read character
bytes imediatelly, cache them until the rest of the character is read
from the process output.

BUG=278340
TEST=in crosh (see comment #1 in the bug):
$python
>>> print(u"\u20ac" * 10000)
Verify a series of EURO signs is displayed (without "unknown" characters)

Review URL: https://codereview.chromium.org/261743002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@270799 0039d316-1c4b-4281-b951-d872f2087c98

• chromeos/process_proxy/process_output_watcher.cc[diff]
• chromeos/process_proxy/process_output_watcher.h[diff]
• chromeos/process_proxy/process_output_watcher_unittest.cc[diff]
• chromeos/process_proxy/process_proxy_unittest.cc[diff]