[clang-tools-extra] [clang-tidy] Add modernize-nlohmann-json-explicit-conversions check (PR #126425)

[Aaron Ballman via cfe-commits]

AaronBallman wrote:

> > It feels like instead nlohmann/json should be the one responsible for providing a "migration path" / tooling for these types of deprecated features.

+1

> > Thus I would vote for keeping clang-tidy as a general tool for C++/STL only.
> 
> N.B. we already have `abseil-`, `boost-`, `cppcoreguidelines-`, `fuschia-`, `linuxkernel-`, and `zircon-`.

`boost` is a well-known third-party library that's often a staging area for STL features; I think they fit in with the STL rather closely.
`abseil`, `cppcoreguidelines`, `cert` are well-known coding standards, which is clang-tidy's niche.
`fuschia` and `zircon` were strongly tied to the LLVM community through Google, but I don't know that I would support adding them if they were proposed today.
`linuxkernel` is another one I don't know that I'd support; it's not that it's a niche project, but more that I don't understand what the value is to the general user base. I think the Linux kernel community could maintain this module (as plugins, even) given that they're the primary beneficiary.

That said, I don't think we've ever had a *policy* about this sort of thing. If we have a consensus opinion forming, we should probably write an RFC and propose some docs to see if the rest of the community buys into it.

https://github.com/llvm/llvm-project/pull/126425