Restructure icu_provider files

[robertbastian]

These are some moves that I've been itching to do before #1576

• resource.rs gets dissolved as it doesn't make sense as a module name after Implement naming convention for "data provider" #1576
  • ResourceOptions into request.rs (new)
  • ResourceKey, ResourceKeyHash into key.rs (new)
• data_provider.rs, a 900-line file, gets split
  • DynProvider and ResourceProvider stay
  • DataRequest and DataRequestMetadata into request.rs (new)
  • DataResponse, DataResponseMetadata, DataPayload, RcWrap into response.rs (new)
  • data_provider/tests.rs are added here
• marker/
  • impls.rs deleted, CowStrMarker usages replaced by HelloWorldV1Marker
  • mod.rs hoisted up as marker.rs

[sffc]

I don't think this is a significant improvement over the status quo, but it's fine

I think the best way to lay these out would be to have one file for each type. ResourceKey, ResourceOptions, DataRequest, DataResponse, and DataPayload all in their own files.

[sffc]

Nit: It does not make sense to have this in hello_world because it is not used by hello_world. I disagree that this is a "toy" implementation; it could very well be used by data structs that contain only a string.

[robertbastian]

It feels like one though. It's very easy for clients to write their own CowStrMarker, and do we actually want to provide a non-ResourceMarker DataMarker as a stable type? Also, why just CowStrMarker and not CowBytesMarker?

I'd also be happy with removing it.

[sffc]

I won't be heartbroken if you remove it, but anywhere it's used, we should migrate to something else. It is mostly used in docs tests. You can change those to something else like HelloWorldV1 if you prefer, or by re-defining CowStrMarker at each call site. However, namespacing it in hello_world is not correct.

[sffc]

Thought: CowStrMarker has convenience functions like DataPayload::<CowStrMarker>::from_static_str. You could achieve mostly the same goal by adding DataPayload::<HelloWorldV1Marker>::from_static_str or HelloWorldV1::from_static_str("...").into_data_payload()

[codecov-commenter]

Codecov Report

Merging #2168 (5d8b2a8) into main (5fa8537) will increase coverage by 0.01%.
The diff coverage is 89.60%.

@@            Coverage Diff             @@
##             main    #2168      +/-   ##
==========================================
+ Coverage   76.60%   76.61%   +0.01%     
==========================================
  Files         400      400              
  Lines       32578    32565      -13     
==========================================
- Hits        24955    24950       -5     
+ Misses       7623     7615       -8     

Impacted Files	Coverage Δ
provider/core/src/any.rs	78.49% <ø> (ø)
provider/core/src/dynutil.rs	84.61% <ø> (ø)
provider/core/src/lib.rs	100.00% <ø> (ø)
provider/core/src/key.rs	87.20% <78.12%> (ø)
provider/core/src/request.rs	89.47% <89.47%> (ø)
provider/core/src/data_provider.rs	88.80% <89.68%> (+0.34%)	⬆️
provider/core/src/response.rs	93.40% <93.40%> (ø)
provider/core/src/hello_world.rs	94.52% <100.00%> (+0.23%)	⬆️
components/locid/src/parser/mod.rs	91.11% <0.00%> (-2.23%)	⬇️
... and 1 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 5fa8537...5d8b2a8. Read the comment docs.

[sffc]

LGTM except for the location of CowStrMarker