fix generic param substitution in templates

[metagn]

fixes #13527, fixes #17240, fixes #6340, fixes #20033, fixes #19576, fixes #19076

When a generic template is called, for the generic parameter in the template body to be substituted with the matching generic parameter, before evaluation the matched generic types are added to the call node as extra arguments. These nodes are nkSym nodes of the original generic parameter symbol, with the type of the symbol and thus the node being the matched type. This is irregular in normal code, and when the template tries to use it, it gets treated as some value with the type instead of the type itself.

To prevent this, remove the type field from these nodes which will make regular code semcheck them and provide them with the appropriate type (similar to #20315). Different places in the compiler treat these symbols differently so there is no uniform computation we can do ahead of time.

[metagn]

PR to fix nimPNG which relied on the bug: jangko/nimPNG#83 (merged)

PR to fix nim-lang NESM (seemingly this is the one CI uses): nim-lang/NESM#1

And original NESM: xomachine/NESM#22

preemptive NESM patch description

NESM is really bothersome to patch, the problem is here:

https://github.com/nim-lang/NESM/blob/0861da16e59974eb5c04ee8777adbcf3f8464d75/nesm.nim#L98-L99

The S argument provided to the macro used to have an S type, now it has a type S type, which the macro has to recognize and skip. This also needs to work with and without the bugfix. Not to mention what I linked is the nim-lang org fork and not the original nesm repo, when a PR would ideally go to both as the entire library seems to rely on that template. Just an annoying situation.

[metagn]

It seems like original NESM has been inactive for a long time now, would be nice if someone could merge the patch into the nim-lang repo (nim-lang/NESM#1) for the time being.

[metagn]

Sorry, the PR to NESM was not enough, made a new one for the nim-lang repo, locally tested this time: nim-lang/NESM#2 and updated the one for the original repo

[metagn]

CI finally passes

[github-actions]

Thanks for your hard work on this PR!
The lines below are statistics of the Nim compiler built from 1cc4d3f

Hint: mm: orc; opt: speed; options: -d:release
169329 lines; 12.760s; 610.566MiB peakmem