LWG4289 Seed sequence is overspecified

[jwakely]

LWG would like LEWG (and maybe SG6) to look at this issue which seems like a design change:

https://cplusplus.github.io/LWG/issue4289

We don't think it can break anything in the standard library, because it only relaxes requirements. This could affect user code which makes use of types meeting the seed sequence requirements and so depends on the APIs proposed for removal. Such code could just document that it depends on the C++11 (or C++23) definition of seed sequence, or document that it needs a seed sequence that meets additional requirements such as default constructibility. Such code would continue to work with all types that do meet those additional requirements.

"Seeding with true sources of randomness is likely to become more important as we look into more ways to mitigate securing exploits in the coming years, so let’s not accidentally make life harder for ourselves than necessary."