[kernel 3d/n] Misc ChainstateManager::Options fixups

[dongcarl]

This is part of the libbitcoinkernel project: #24303, https://github.com/bitcoin/bitcoin/projects/18

This PR is NOT dependent on any other PRs.

---

Places ChainstateManager::Options into the kernel:: namespace and use designated initializers for construction.

[dongcarl]

@MarcoFalke wondering if you know why the Desig usages are not working here? https://cirrus-ci.com/task/6347551300386816?logs=ci#L2321

init.cpp does #include <util/designator.h>

[maflcko]

You'll need to rebase 😅

[dongcarl]

Intermittent rpc_net issue strikes again! https://cirrus-ci.com/task/6461219992240128?logs=ci#L4516

(restarting CI job)

[ryanofsky]

Code review ACK ce8b0f9

I didn't realize Desig(field_name) compiled to nothing on certain platforms. That seems like it was dangerous because code might appear to be initializing one field, but actually be initializing another if fields were not in order. Unless maybe field order was enforced through CI. Anyway glad that is gone now.

[maflcko]

Yes, it was enforced on every platform except msvc.

[ryanofsky]

Yes, it was enforced on every platform except msvc.

Thanks, I didn't realize out of order designated initializers were actually forbidden by the language.

[ryanofsky]

Thanks, I didn't realize out of order designated initializers were actually forbidden by the language.

Though if skipping a field was allowed this still could have been bad.

[maflcko]

Skipping is allowed, and seems no worse than omission of "trailing" fields

[ryanofsky]

Wouldn't this assign (1, 2, 0) on msvc and (1, 0, 2) on other platforms?

struct S {
   int a;
   int b;
   int c;
};

S s{
  Desig(a) 1,
  Desig(c) 2,
};

[maflcko]

Yes (oops)