Documentation on Collector behaviour needs clarification

[Spacerat]

After grappling with some speed issues in develop mode, I realised that Collector.collect() was being run for every pipeline tag in my project. Eventually I narrowed my issue down to this:

https://github.com/jazzband/django-pipeline/blob/master/pipeline/collector.py#L31-L32

request was always None for me, and so the solution was to add 'django.core.context_processors.request' to my project's TEMPLATES['OPTIONS']['context_processors'] option which gave me a roughly 6x speedup.

I imagine it's probably enough to just add something to the documentation which instructs users to add the context processor if they intend to use multiple tags, although I'm not sure where. Installation→Recommendations could work, or maybe near the top of the Usage or Configuration sections.