MSE: Test addSourceBuffer and changeType relaxed type strictness and allowance for implicit changeType

[chromium-wpt-export-bot]

In preparation for Chrome's relaxation of addSourceBuffer and
changeType codec specificity within the mime type parameter, this
change adds new new tests. It also updates the Blink test expectations
for these new tests to be failures. Later changes will add Chrome's
implementation of the relaxed strictness, and will also remove the
failure expectations.

external/wpt/media-source/mediasource-changetype-play-without-codecs-parameter:
Tests successful addSourceBuffer, changeType, and codec switching
without using any codecs parameter in types passed to addSourceBuffer
and changeType for pairs of test media that are of same media class
(audio or video) and are single track. This test is very similar to
mediasource-changetype-play.html, just with less specific parameters
to addSourceBuffer and changeType. This file is kept separate from
mediasource-changetype-play.html to help identify implementations
(like Chrome at the time of this change) that require more specific
parameters to those methods.

external/wpt/media-source/mediasource-changetype-play-implicit.html:
Tests successful codec switching without using changeType for test
media of the same bytestream format, separately for audio-only and
video-only pairs of media. Also includes a set of the same tests where
the initial addSourceBuffer uses only the (relaxed) mime type/subtype
without any codecs parameters, for any pairs that included codecs
parameters in their full types.
Note: In Chrome, only 1 pair of actually different media files is
supported across the test media: webm vp8 and vp9. We can add more
test media later if greater coverage is desired.

external/wpt/media-source/mediasource-changetype-play-negative.html:
Tests of various explicit, implicit, strict and relaxed scenarios
which should fail.

Main MSE spec issue for this:
w3c/media-source#161
Related codec-switching MSE spec issue:
w3c/media-source#155
These new tests exercise related scenarios to the new non-normative
guidance in the codec-switching incubation spec branch in WICG
(https://github.com/WICG/media-source/tree/codec-switching),
clarifying that implementations can be relaxed in addSourceBuffer
and changeType codec-specificity, but not isTypeSupported; apps
should still provide as much type specifics as they can to achieve
earlier confidence in support or lack thereof, and to avoid issues
with implementations (like Chrome at the time of this change) that
require more specific parameters to those methods.

BUG=535738

Change-Id: I20fd477b2429ef94ee70bf57ed8c18543774da93
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/1663349
Reviewed-by: Dan Sanders <sandersd@chromium.org>
Commit-Queue: Matthew Wolenetz <wolenetz@chromium.org>
Cr-Commit-Position: refs/heads/master@{#672316}

[wpt-pr-bot]

Already reviewed downstream.

[lukebjerring]

Unstable results

Test	Subtest	Results	Messages
/media-source/mediasource-changetype-play-negative.html		ERROR: 7/10, OK: 3/10

[Hexcles]

The original Chromium commit introduced a new flaky test on Chrome, caused by an context-less assertion introduced by the util script; I pushed a commit to fix that: 1a67965 , which was reviewed by @jgraham (thanks!).

Unfortunately the change in the util script "affects" all tests in media-source and exposes some preexisting flakiness on Firefox (those tests aren't really affected by this PR).

[wolenetz]

Thank you, @Hexcles and @jgraham.