Add --waves argument to generate waveforms for GHDL and NVC

[oscargus]

Closes #1099
Closes #1042

The approach here is to pass --waves if a waveform file should be generated. That is clearer than the "passing --gtkwave-fmt "-approach earlier used.

There is a deprecation warning when passing only --viewer-fmt/--gtkwave-fmt for GHDL, but it still generates the file.

I also added a new alias --waves-fmt for consistency.

(I guess that one can also support other simulators generating waveform files. At least for ModelSim/Questa it is possible to generate VCD-files with a few .do-file commands and I guess the same holds for many other simulators.)

[JHertz5]

I see that this PR has been open for quite a while. I just want to chip in and say that I think this is a great addition, and it would be very useful to me if merged.

[creiter64]

I also agree that this is a great addition as it allows to reload the waveform in GTKWave for different test runs.
@LarsAsplund could you also please take a look at this proposal? I think it makes sense to separate generating the waveforms from opening a waveform viewer. This would also integrate well with fusesoc in my workflow.

[LarsAsplund]

@oscargus I like this but have two comments:

1. I prefer --wave over --waves to indicate a feature rather than what we physically produce considering that we produce a single file which may also contain a single wave. Also, "waves format" sounds weird.
2. If I use -g, I will see a waveform but if I use --wave(s), I don't get a file unless I specify the format. I think we should use the same default for both cases.

[oscargus]

I've updated to --wave and --wave-fmt.

You do not have to specify --wave-fmt. The thing is that it is not enough to only pass --wave-fmt to get a wave form file. You will also have to pass --wave. That is something one can think about, but it is at least consistent with the g flag. (The defaults are the same since --wave-fmt is just an alias for the others so one can actually pass --wave --gtkwave-fmt=fst.)