mojo_run: support parallel Android runs of mojo_run.

mojo_run: support parallel Android runs of mojo_run.

This patch adds a `--free-host-ports` flag to devtools that allows to
run dev servers on system-allocated ports (so that we can have parallel
runs) while still using fixed ports on the device (so that the caching
still works).

Running with `--free-host-ports` makes `adb_remote_setup` unable to work
correctly, which is why this is being introduced behind a flag.

Fixes #470.

R=qsr@chromium.org

Committed: https://chromium.googlesource.com/external/mojo/+/eec06de72752abcae761e709de28e9eebd3fa52a

[ppi, 2015-10-13 21:55:37]

Hi Ben,

ptal.

[qsr, 2015-10-14 07:35:39]

LGTM if the response to my question is no.

https://codereview.chromium.org/1397193005/diff/1/mojo/devtools/common/devtoo...
File mojo/devtools/common/devtoolslib/shell_arguments.py (left):

https://codereview.chromium.org/1397193005/diff/1/mojo/devtools/common/devtoo...
mojo/devtools/common/devtoolslib/shell_arguments.py:105: def
configure_local_origin(shell, local_dir, fixed_port=True):
Was someone using this? If no, no problem, but otherwise you changed the
behavior => previously, fixed_port=False would have changed the port on the
device too...

[ppi, 2015-10-14 16:53:20]

https://codereview.chromium.org/1397193005/diff/1/mojo/devtools/common/devtoo...
File mojo/devtools/common/devtoolslib/shell_arguments.py (left):

https://codereview.chromium.org/1397193005/diff/1/mojo/devtools/common/devtoo...
mojo/devtools/common/devtoolslib/shell_arguments.py:105: def
configure_local_origin(shell, local_dir, fixed_port=True):
On 2015/10/14 07:35:39, qsr wrote:
> Was someone using this? If no, no problem, but otherwise you changed the
> behavior => previously, fixed_port=False would have changed the port on the
> device too...

According to git grep, no-one was using this.

[ppi, 2015-10-14 16:54:04]

Committed patchset #1 (id:1) manually as
eec06de72752abcae761e709de28e9eebd3fa52a (presubmit successful).