Switch to double for time calculations using playback rate.

Multiplying an int64 against a float causes the float type to prevail which does not have enough precision for accurate time value computations.

BUG= 475964

TBR=lcwu1, gunsch

Review URL: https://codereview.chromium.org/1094783002

Cr-Commit-Position: refs/heads/master@{#326269}
48 files changed