Add a Worker configuration option to configure the Linux out-of-memory
behaviour. Add `oom_score_adjust=500` to `flamenco-worker.yaml` to increase
the chance that Blender gets killed when the machine runs out of memory,
instead of Flamenco Worker itself.
Add a little intro that explains that there are multiple config files. And
add a section that explains that the worker-specific files are not to be
edited, just deleted.
The example on the project website is incorrect:
```
task_types: blender, ffmpeg, file-management, misc
```
It results in an error:
```
loading configuration: yaml: unmarshal errors:\n line 2:
cannot unmarshal ||str "blender..." into []string
```
The fix seems to be adding square brackets around the list of comma
delineated tasks:
```
task_types: [blender, ffmpeg, file-management, misc]
```
This results in output with the confirmation that
`taskType=["blender","ffmpeg","file-management","misc"]`
The concept of "job types" is more important. The fact that they are
defined by scripts is secondary. I've also added more documentation about
the settings & task types.
The flamenco.io website moves to flamenco.blender.org!
This commit updates the Makefile as follows:
- Rename 'site' to 'project-website', so the new command to deploy is
'make project-website'
- Move the website directory from 'flamenco-io-site' to
'project-website'
- Update the rsync command do reflect the new deployment destination
