Add laser parameter output for --dump-metadata

[max-lehman14]

This pull request adds laser parameter output in the BaseParam.def and GaussianPulse.def files, based on the new parameter output feature from @chillenzer .

In the picongpu-metadata.json file the standard laser parameters from incidentField.param are displayed with units, for example:

"W0": {
    "unit": "m",
    "value": 4.246609082647506e-06
}

or

"direction": {
    "unit": "none",
    "value": [
      0.0,
      1.0,
      0.0
    ]
},

The next things on my list are:

• adding README [EDIT/REMOVED by @PrometheusPi]

• invest deeper in the problem, that Laguerremodes and -phases in the incidentField.param file are not correctly displayed if there are several Laguerremodes (seems like its falling back to the default 0th Laguerre mode for a standard Gaussian)

[chillenzer]

Without having looked at the content I can already see that this file didn't change, so almost certainly you didn't run the tests (as the failing CI is also telling you). Please update said file and check that the tests and examples all compile and pass.

[chillenzer]

@PrometheusPi @max-lehman14 What's the status here?

[PrometheusPi]

@chillenzer I will take over this PR

[PrometheusPi]

This still does not output anything else than the default parameters. And there should be a bug in the Gaussian code that is actually never causing any compile time issues.

[PrometheusPi]

@steindev @chillenzer @psychocoderHPC ready for review
I will not add a README since I do not see the benefit of that. If you see any, please let m eknow.

[steindev]

Just one comment.

[steindev]

We typically use T_ as default names for template arguments and not My. I would like to stick to that guideline.

[PrometheusPi]

This was already part of the original implementation from @chillenzer.
I agree with you. @psychocoderHPC should this change go in this PR as well?

[PrometheusPi]

renamed My to T_Self

[chillenzer]

I'm aware of this and I didn't stick to that convention because I see it less as a template parameter and more as a hack/workaround for injecting the correct type.

I'm fine with changing it but if so, we should probably do it consistently everywhere and adopt that as the new convention.

[PrometheusPi]

Interstingly, we get compile errors like:

crp/picongpu/include/picongpu/../picongpu/fields/incidentField/profiles/BaseParam.def:226:63: error: no member named 'TIME_DELAY_SI' in 'picongpu::fields::incidentField::profiles::MyGaussianParam'
  226 |                         result["time_delay"] = {{"value", My::TIME_DELAY_SI}, {"unit", "s"}};

which I did not see when I tested the code on hemera.

[PrometheusPi]

The time delay was actually outputted correctly for my LWFA test - strange

[psychocoderHPC]

MUST BE GetTimeDelay<My>::value

[PrometheusPi]

Thanks - I applied it

[chillenzer]

You've done everything right so far but the CI failures are due to PlaneWaveParam in some examples not inheriting from BaseParam, see e.g. in Bunch and analogously in FoilLCT. This needs to be added and in the best case, we go through the examples and tests by hand and check for such missing inheritance.