[omnibox] Use token sequences for Pedal concept matching

The experimental string-based Pedal concept matching code is replaced
by a custom implementation using token sequences. The overall technique
of erasing found subsequences remains intact but the new implementation
is much faster and facilitates internationalization by supporting
alternative token structures (e.g. in languages that do not divide words
using spaces).

Bug: 893183
Change-Id: I71be5ceee9f73ec9758394ae32b9d27fe9fed0bf
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/1531724
Commit-Queue: Orin Jaworski <orinj@chromium.org>
Reviewed-by: Tommy Li <tommycli@chromium.org>
Cr-Commit-Position: refs/heads/master@{#643490}
9 files changed