Upgrade downstreamPlaceholder to a type with methods

[Ericson2314]

Motivation

This gets us ready for dynamic derivation dependencies (part of RFC 92).

Context

Tracking issue for RFC 92 #6316

Checklist for maintainers

Maintainers: tick if completed or explain if not relevant

• agreed on idea
• agreed on implementation strategy
• tests, as appropriate
  • functional tests - tests/**.sh
  • unit tests - src/*/tests
  • integration tests - tests/nixos/*
• documentation in the manual
• documentation in the internal API docs
• code and comments are self-explanatory
• commit message explains why the change was made
• new feature or incompatible change: updated release notes

Priorities

Add 👍 to pull requests you find important.

[Ericson2314]

OK think I addressed everything.

[roberth]

C++ lgtm. My suggestions to use both file static and class static were incoherent, but you resolved that well.

I assume we will later have a use for this type, and not just the construct + render combination?

Also just to be sure, what's the stability status of these hashes? I would hope they are meant to be stable but a change (if we were to accidentally make one) would rarely be observed by users? Or am I way off? Wouldn't nix eval print them? Is this documented somewhere? Do we have a test that checks the placeholder against a hardcoded example to make sure it doesn't change by accident?

[Ericson2314]

I assume we will later have a use for this type, and not just the construct + render combination?

Yes, once inputDrvs becomes fancier to include inductive stuff.

Also just to be sure, what's the stability status of these hashes?

Yes they will end up in derivations (in memory and drv files, so they shouldn't change). If they do change, trying to build old ca or dynamic-derivation-depending derivations could fail because they no longer get substituted.

We could write some tests, certainly. Don't think there are interesting properties to test, but we could just put in a fixed case of each.

(Similarly we should pin down the hash-modulo stuff at some point too.)

[Ericson2314]

I added some tests. The first I added test prior to the refactor so we can see that behavior is preserved.