TemporaryPipInstallResolver: Prevent pip install from polluting stdout

[jherland]

I noticed this while playing around with --install-deps and --json, and
at the same time piping the JSON output into jq for filtering/colors:
The pip install that is triggered by --install-deps will print status
messages to both stdout and stderr, and since these are not handled by
our subprocess.run() call, they will appear directly on our
stdout/stderr. When piping our stdout to jq this causes jq to get
those status messages (preceding the JSON output) on its stdin, and it
predictably aborts with an error message.

This commit fixes the issue in two ways:

First, we pass --quiet and --disable-pip-version-check to pip install,
to silence some common/unnecesary output. This seems sufficient to make
pip install silent when everything works as expected. However, when
warnings and errors might still be printed in exceptional cases.

Next, we capture the stdout + stderr from pip install and only print
them in case the pip install command failed.

With both of these fixes, we ensure that pip install only makes noise
when it fails, and that noise only appears on our stderr (in the shape
of a warning log message). For example, if I try to --install-deps with
an intentional typo (pyyamll):

  WARNING:fawltydeps.packages:Command failed: ['/run/user/1000/tmpr5l6m94i/bin/pip', 'install', '--no-deps', '--quiet', '--disable-pip-version-check', 'pyyamll', 'scikit-learn']
  WARNING:fawltydeps.packages:Output:
  ERROR: Could not find a version that satisfies the requirement pyyamll (from versions: none)
  ERROR: No matching distribution found for pyyamll

  WARNING:fawltydeps.packages:Failed to install 'pyyamll'
  WARNING:fawltydeps.packages:Output:
  ERROR: Could not find a version that satisfies the requirement pyyamll (from versions: none)
  ERROR: No matching distribution found for pyyamll

  ERROR:fawltydeps.main:Unresolved dependencies: pyyamll
  FawltyDeps is unable to find the above packages with the configured package resolvers. Consider using --pyenv if these packages are already installed somewhere, or --custom-mapping-file to take full control of the package-to-import-names mapping.