Stablize and flip repository_cache_urls_as_default_canonical_id

Introduced in bazelbuild#14268, this flag is very useful for bigger enterprise
context where folks often version bumping dependencies without
remembering to update the SHA256 of the downloaded file, leading to
Bazel picking up older download entries from the repository cache.

As we get more questions about this flag in Slack, marking it as
stable and flip the default seems to be the right move.

src/main/java/com/google/devtools/build/lib/bazel/repository/RepositoryOptions.java

@@ -287,8 +287,9 @@ public Converter() {
   public CheckDirectDepsMode checkDirectDependencies;
 
   @Option(
-      name = "experimental_repository_cache_urls_as_default_canonical_id",
-      defaultValue = "false",
+      name = "repository_cache_urls_as_default_canonical_id",
+      oldName = "experimental_repository_cache_urls_as_default_canonical_id",
+      defaultValue = "true",
       documentationCategory = OptionDocumentationCategory.BAZEL_CLIENT_OPTIONS,
       effectTags = {OptionEffectTag.LOADING_AND_ANALYSIS},
       metadataTags = {OptionMetadataTag.EXPERIMENTAL},

src/test/shell/bazel/bazel_repository_cache_test.sh

@@ -494,18 +494,18 @@ EOF
     || echo "Expected fetch to succeed"
 }
 
-function test_experimental_repository_cache_urls_as_default_canonical_id() {
+function test_repository_cache_urls_as_default_canonical_id() {
   setup_repository
 
   bazel fetch --repository_cache="$repo_cache_dir" \
-        --experimental_repository_cache_urls_as_default_canonical_id \
+        --repository_cache_urls_as_default_canonical_id \
         //zoo:breeding-program >& $TEST_log \
     || echo "Expected fetch to succeed"
 
   shutdown_server
 
   bazel fetch --repository_cache="$repo_cache_dir" \
-        --experimental_repository_cache_urls_as_default_canonical_id \
+        --repository_cache_urls_as_default_canonical_id \
         //zoo:breeding-program >& $TEST_log \
     || echo "Expected fetch to succeed"
 
@@ -524,7 +524,7 @@ EOF
 
   # As repository cache key should depend on urls, we expect fetching to fail now.
   bazel fetch --repository_cache="$repo_cache_dir" \
-        --experimental_repository_cache_urls_as_default_canonical_id \
+        --repository_cache_urls_as_default_canonical_id \
         //zoo:breeding-program >& $TEST_log \
     && fail "expected failure" || :
 }