First example should probably say PipelineStorage

Fix incorrect PIPELINE_JS_COMPRESSOR setting in documentation

Adding a new company on the documentation (fixes #309).

The current code inserts bytes into a string. In Python 3 those are different types, causing the URI to look like:

    url("data:image/png;charset=utf-8;base64,b'iVBORw0KGgoAAAANSUhEUgAAAA...'")

Because Base64 data is always ASCII it is safe to decode it as such, and insert the resulting string, which causes the correct output:

    url("data:image/png;charset=utf-8;base64,iVBORw0KGgoAAAANSUhEUgAAAA...")

First example in documentation should refer to ordinary PipelineStorage rather than PipelineCachedStorage, which is covered in the "Cache-busting" section below.

Do not rewrite data URIs

Fixed spelling of my last name

If multiprocessing and concurrent unspported, render linear compilation.

Use generators and add a gzip support.

Signed-off-by: Timothée Peignier <timothee.peignier@tryphon.org>

Avoid listing all files for every finder for finding a file. This is
actually a bottleneck when the directory hierarchy is nested or the
number of files served in DEBUG mode is big. Since this flow is done in
every request, it slows down page loading.

In production environments where templates are usually cached performance
doesn't get hit by this.

Minor performance improvement, avoid memory relocation every time we
iterate and do a string concatenation. Instead add results to a list and
join them at the end.

References issue #282.

I added an import for `mark_safe` and wrapped the JavaScript / CSS
urls in it. All tests passed.

prevent pipeline from dying on Closure Compiler warnings

Coffeescript compiler now accepts .litcoffee and .coffee files.

fixed issue #271