feat(spanner): split GAPIC from the spanner handwritten package

[quirogas]

Decouple the handwritten Spanner wrapper from the auto-generated GAPIC client layer by importing a standalone version of the GAPIC library. Standalone -api packages are now generated and maintained independently.

• Removed embedded GAPIC generated code (src/v1/) and OwlBot files (owlbot.py, .OwlBot.yaml).
• Added dependency on standalone @google-cloud/spanner-api (^0.2.0).
• Re-exported v1, top-level sub-clients (SpannerClient, DatabaseAdminClient, InstanceAdminClient, SpannerExecutorProxyClient), and protos from @google-cloud/spanner- api.
• Resolved protobufjs status definition loading in mock server test utilities.

Internal: b/531797111

[gemini-code-assist]

Code Review

This pull request replaces local proto files and generated clients with the external '@google-cloud/spanner-api' package, updating import paths across the codebase, removing obsolete generator scripts, and updating package dependencies. The review feedback suggests importing 'protos' directly from the public entry point of '@google-cloud/spanner-api' instead of deep-importing, simplifying the client re-exports in 'src/index.ts', and using a cleaner direct require path for 'status.json' from 'google-gax' in the mock server tests.

[gemini-code-assist]

Instead of deep-importing from the internal build/ directory of @google-cloud/spanner-api, you can import protos directly from the public entry point of the package. This is safer, more maintainable, and avoids potential module resolution issues if the package's internal structure changes or if strict package exports are enforced.

Suggested change

	import * as protos from '@google-cloud/spanner-api/build/protos/protos';
	import {protos} from '@google-cloud/spanner-api';

[gemini-code-assist]

These lines are highly verbose and redundant. In TypeScript, you can directly re-export both the values and types of these clients from @google-cloud/spanner-api in a single statement. Additionally, this ensures that SpannerExecutorProxyClient is also re-exported at the top level, as mentioned in the pull request description.

Suggested change

	const DatabaseAdminClient = v1.DatabaseAdminClient;
	type DatabaseAdminClient = v1.DatabaseAdminClient;
	const InstanceAdminClient = v1.InstanceAdminClient;
	type InstanceAdminClient = v1.InstanceAdminClient;
	const SpannerClient = v1.SpannerClient;
	type SpannerClient = v1.SpannerClient;
	export { DatabaseAdminClient, InstanceAdminClient, SpannerClient };
	export { DatabaseAdminClient, InstanceAdminClient, SpannerClient, SpannerExecutorProxyClient } from '@google-cloud/spanner-api';

[gemini-code-assist]

Instead of using a fragile combination of require.resolve, path.dirname, and path.join to locate status.json inside google-gax, you can require it directly as 'google-gax/protos/status.json'. This is much cleaner, more readable, and less prone to breaking if the internal file structure of google-gax changes.

Suggested change

	const protobufjs = require('protobufjs');
	const statusProtoJson = require(path.join(path.dirname(require.resolve('google-gax')), '../protos/status.json'));
	const RetryInfo: any = protobufjs.Root.fromJSON(statusProtoJson).lookupType('google.rpc.RetryInfo');
	const protobufjs = require('protobufjs');
	const statusProtoJson = require('google-gax/protos/status.json');
	const RetryInfo: any = protobufjs.Root.fromJSON(statusProtoJson).lookupType('google.rpc.RetryInfo');

[feywind]

Closing this one as a dup for #8927, integrated some bits of it!