Workaround for #34

[arnetheduck]

This PR adds a workaround for the case where a global symbol is matched instead of the local error/value template in valueOr and friends, when valueOr is being used in a generic context.

Two options are added:

• when using Nim version supporting the experimental genericsOpenSym feature, we use it
• when not, a macro tries to replace all accesses in the given body with the correct symbol - this matching is done by name and some minimal heuristics which potentially could be wrong - the upstream solution is obviously preferable

Both solutions can be disabled via compile-time options to retain the old behavior of matching the global symbol.

See also nim-lang/Nim#22605

thanks to @Araq for the macro hack and @metagn for the language fix!

[etan-status]

Related: status-im/nim-stew#174

[etan-status]

How well does this work when passing error as a kv param to a chronicles log? will the hack become user visible?

[etan-status]

Also, how big is the desire to have any workaround within the status repos, if we rename the injected name to err instead of error? How about resultErr or resError?

[arnetheduck]

How well does this work when passing error as a kv param to a chronicles log? will the hack become user visible?

works fine - this is actually where the problem manifests in our code: an error enum gets logged as "error" instead of the actual error we want, due to the automatic string conversion that logging does