Ghostscript (gstoraster): Introduce cupsHalftoneType dithering algorithms

[ValdikSS]

Add new dithering (halftone) algorithms in addition to the default Ghostscript one.
Controlled either with halftone-type job option or cupsHalftoneType PPD option.

• Stochastic algotirhm is implemented in stocht.ps Ghostscript library, just include it if cupsHalftoneType is set to yes/true/on/stochastic.
• Foo2zjs algorithm is a PostScript code taken from foo2zjs-pstops file (don't know the name of the algorithm)
• Threshold algorithm

Default algorithm

Stochastic algorithm

Foo2zjs algorithm

[ValdikSS]

Is it fine to introduce new prefixed PPD option? Some proprietary filters already use Halftone and HalftoneType, so I decided to go with cupsHalftoneType, even if it only applies to Ghostscript-based filter.

cups-filters 1.x: OpenPrinting/cups-filters#642

[zdohnal]

The code looks ok on syntax side of things, there are some space/tab issues against applied code style.

Regarding functionality I can't say, since I don't fully understand the best practices and it would be great if it was reviewed by an additional person.

[zdohnal]

Either way, thank you for the PR, @ValdikSS !

Ad new prefixed option - I would say it is fine, but probably @michaelrsweet should know more.

[zdohnal]

one line scopes should not be enveloped by brackets, otherwise it looks great!

[zdohnal]

Looks great, thank you!

@michaelrsweet / @tillkamppeter can you review this from functioanity POV (meaning whether halftone implementation is correct)? I have check foo2zjs code and the Postscript code here is matching this PR, bi-level looks feasible with my little knowledge of Postscript, stocht.ps from ghostscript is accessible on machines which includes the script, so LGTM.

[tillkamppeter]

Principally, this code looks OK to me.

Some remarks:

1. When not using the new options, does one get the same results as before?
2. Will the halftoning also work if the input into Ghostscript is PDF?
3. As long as the new halftoming options are not represented in print dialogs, nearly no user will use them, as they do not read the documentation of libcupsfilter
4. The feature will only be available if the job goes through Ghostscript. Jobs with a raster or image format as input for example will not be affected by the new halftoning settings.

[ValdikSS]

When not using the new options, does one get the same results as before?

Yes.

Will the halftoning also work if the input into Ghostscript is PDF?

Yes, I've tested on PDF files only.

As long as the new halftoming options are not represented in print dialogs, nearly no user will use them, as they do not read the documentation of libcupsfilter

Indeed, proper integration would be much better. Any suggestions?

Jobs with a raster or image format as input for example will not be affected by the new halftoning settings.

poppler already has bi-level halftoning, and its default is much better than ghostscript's default.