fix: Fix grpc schema hack in flight integration test

[alamb]

Which issue does this PR close?

Closes #1398

Rationale for this change

Certain flight integration tests are failing with rust

We think something about the gRPC library has changed recently to be more strict, but I never found a good way to verify this.

What changes are included in this PR?

Change some workaround code in the "scheme" passed to gRPC to use http:// rather thangrpc:// #1398 (comment) as suggested by @lidavidm ❤️ ❤️

Are there any user-facing changes?

No this is a test only change

[codecov-commenter]

Codecov Report

Merging #1402 (c8fb66d) into master (afcf304) will decrease coverage by 0.00%.
The diff coverage is 0.00%.

❗ Current head c8fb66d differs from pull request most recent head 6132b3b. Consider uploading reports for the commit 6132b3b to get more accurate results

@@            Coverage Diff             @@
##           master    #1402      +/-   ##
==========================================
- Coverage   83.10%   83.10%   -0.01%     
==========================================
  Files         181      181              
  Lines       53244    53244              
==========================================
- Hits        44249    44248       -1     
- Misses       8995     8996       +1     

Impacted Files	Coverage Δ
...ng/src/flight_client_scenarios/integration_test.rs	0.00% <0.00%> (ø)
parquet_derive/src/parquet_field.rs	65.98% <0.00%> (-0.46%)	⬇️
arrow/src/array/transform/mod.rs	86.43% <0.00%> (+0.11%)	⬆️

---

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 afcf304...6132b3b. Read the comment docs.

[alamb]

🎉

[lidavidm]

Just some context on this bit: Flight's Location is never actually directly used as a gRPC URI. The C++ and Java implementations take apart and reconstruct the URI, so, grpc+tcp was never menat to be passed to gRPC. You can see here: https://github.com/apache/arrow/blob/4ef95eb89f9202dfcd9017633cf55671d56e337f/cpp/src/arrow/flight/client.cc#L935-L939

I do wonder if gRPC got more strict. The last passing run used gRPC 1.43.2, the first failing run used gRPC 1.44.0. By my read, this is the responsible commit, first seen in gRPC 1.44.0: grpc/grpc@0deb64d

It adds the error message to hpack_parser.cc (see ReportMetadataParseError) and crucially adds new validation to metadata_batch.h (see HttpSchemeMetadata). So I guess this was new validation added in v1.44.0.