Add an example for accessing the resolved locale

[sffc]

In relation to #3906 and #58

[sffc]

@robertbastian

---- src/lib.rs - data_provider (line 327) stdout ----
error: doctest failed, to rerun pass `-p tutorials-test --doc`
error[E0425]: cannot find function `any` in crate `icu_testdata`
  --> src/lib.rs:356:26
   |
31 |     inner: icu_testdata::any(),
   |                          ^^^ not found in `icu_testdata`
   |
error: aborting due to previous error

For more information about this error, try `rustc --explain E0[425](https://github.com/unicode-org/icu4x/actions/runs/5958142030/job/16161946606?pr=3928#step:9:426)`.
Couldn't compile the test.

failures:
    src/lib.rs - data_provider (line 327)

The local icu_testdata does not export an any() function unless you have the icu_locid_transform feature enabled. This is a breaking change. In 1.2, the required functionality lived in icu_provider_adapters which is non-optional. So I guess we should either make icu_locid_transform be non-optional, or do the crate split now (splitting icu_locid_transform into two crates, one more core than the other).

[robertbastian]

The local icu_testdata does not export an any() function unless you have the icu_locid_transform feature enabled. This is a breaking change. In 1.2, the required functionality lived in icu_provider_adapters which is non-optional. So I guess we should either make icu_locid_transform be non-optional, or do the crate split now (splitting icu_locid_transform into two crates, one more core than the other).

icu_locid_transform is a default feature, semver doesn't apply with default-features = false

[robertbastian]

I guess this could be take instead of clone? The FixedDecimalFormatter should not need the locale.

[robertbastian]

I'd really prefer if this would take. This would demonstrate that the locale is not actually used later, and it's only there to allow writing this specific wrapper.

[robertbastian]

The tutorials test is failing because the legacy_testdata crate only exposes what is needed for tutorials, and any() hasn't been so far. In fact, could you use another provider for this demo? I'd rather not add a testdata tutorial, and I'd also rather not add an AnyProvider tutorial.

[sffc]

The tutorials test is failing because the legacy_testdata crate only exposes what is needed for tutorials, and any() hasn't been so far.

OK, so I should just go to the tutorials test Cargo.toml and enable the required feature.

In fact, could you use another provider for this demo?

Suggestion?

I'd rather not add a testdata tutorial, and I'd also rather not add an AnyProvider tutorial.

I don't know how to do this in a scalable way without using AnyProvider. And we already have an AnyProvider tutorial.

In general, this is a power user feature, so I'm okay forcing users who need this feature to use a custom manual data build.

[robertbastian]

OK, so I should just go to the tutorials test Cargo.toml and enable the required feature.

You have to add the actual function, it's a custom lib.rs.

We're deprecating icu_testdata in 1.3, there should be some way to demonstrate this without it. I don't want icu_testdata in tutorials after 1.3.

[sffc]

We're deprecating icu_testdata in 1.3, there should be some way to demonstrate this without it. I don't want icu_testdata in tutorials after 1.3.

I hear you. I'm second-guessing that decision to deprecate now. We still use icu_testdata as a way to get a blob or any provider in tests, which is not a deprecated piece of functionality.

[sffc]

I guess I could just use HelloWorldProvider.

[sffc]

I think I need to add fallbacker data to the tutorials testdata crate

error[E0277]: the trait bound `UnstableDataProvider: DataProvider<LocaleFallbackLikelySubtagsV1Marker>` is not satisfied
   --> src/lib.rs:361:44
    |
error: doctest failed, to rerun pass `-p tutorials-test --doc`
36  |         LocaleFallbacker::try_new_unstable(&icu_testdata::unstable()).unwrap(),
    |         ---------------------------------- ^^^^^^^^^^^^^^^^^^^^^^^^^ the trait `DataProvider<LocaleFallbackLikelySubtagsV1Marker>` is not implemented for `UnstableDataProvider`
    |         |
    |         required by a bound introduced by this call