fix: URL encode explicit routing values

[dbolduc]

Part of the work for #12232

Update the explicit routing matchers to url encode the values in the routing pairs.

Update the part of testing_util::ValidateMetadataFixture that computes the expected routing headers.

Then legacy bigtable also uses that testing helper, so we must update it in this PR. I separated the bigtable stuff into its own commit (hopefully) for clarity.

---

This change is 

[codecov]

Codecov Report

Patch coverage: 100.00% and project coverage change: -0.01% ⚠️

Comparison is base (d67bcb7) 93.61% compared to head (f7e2e28) 93.60%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #12447      +/-   ##
==========================================
- Coverage   93.61%   93.60%   -0.01%     
==========================================
  Files        2036     2036              
  Lines      180316   180340      +24     
==========================================
+ Hits       168794   168810      +16     
- Misses      11522    11530       +8     

Files Changed	Coverage Δ
...sts/golden_kitchen_sink_metadata_decorator_test.cc	99.61% <100.00%> (ø)
...olden_kitchen_sink_rest_metadata_decorator_test.cc	100.00% <100.00%> (ø)
...golden_thing_admin_rest_metadata_decorator_test.cc	100.00% <100.00%> (ø)
.../cloud/bigtable/internal/legacy_row_reader_test.cc	100.00% <100.00%> (ø)
google/cloud/bigtable/metadata_update_policy.cc	100.00% <100.00%> (ø)
...ogle/cloud/bigtable/metadata_update_policy_test.cc	100.00% <100.00%> (ø)
google/cloud/internal/routing_matcher.h	100.00% <100.00%> (ø)
google/cloud/internal/routing_matcher_test.cc	95.38% <100.00%> (+1.90%)	⬆️
google/cloud/testing_util/validate_metadata.cc	92.81% <100.00%> (ø)

... and 6 files with indirect coverage changes

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[devbww]

Nit: I was going to suggest using UrlEncode() to construct these expected strings, but them I thought, "nah, it's fine putting exactly what is expected." But then I saw that you do use UrlEncode() in other tests. Perhaps those latter cases are just because you have the unencoded string for other purposes? Anyway, perhaps worthy of a short ponder.

[dbolduc]

Hm, so I didn't consider that. Now that I am considering it, I think I ever so slightly prefer the code the way it is.

We do not URL encode the =, so it would look like:

auto expected1 = "table_location=" + internal::UrlEncode("instances/instance_bar");

or really...

auto expected1 = internal::UrlEncode("table_location") + "=" + internal::UrlEncode("instances/instance_bar");

.... meh