feat: regenerate module for networksecurity

[blakeli0]

Command used:

python3 generation/new_client_hermetic_build/add-new-client-config.py add-new-library
--api-shortname="networksecurity"
--name-pretty="Network Security API"
--proto-path="google/cloud/networksecurity/v1"
--product-docs="https://cloud.google.com/products/networking"
--api-description="The Network Security API provides resources for configuring and managing network security."
--googleapis-committish="29c6482ccb75b302c33665d8c7d7a38d3698be53"
--library-name="network-security"
--release-level="stable"

[blakeli0]

PiperOrigin-RevId: 919268561

[gemini-code-assist]

Code Review

This pull request introduces a mechanism to refresh transport channels upon encountering an UnauthenticatedException when running in a specific environment (isMwlidEnvironment), along with rate-limiting for channel pool refreshes. Feedback on the changes highlights a critical bug in GrpcCallContext where a renamed parameter was not updated, leading to a potential NullPointerException. Additionally, the rate-limiting logic in ChannelPool could fail if System.nanoTime() returns a negative value, and querying the environment variable on every retry check in ApiResultRetryAlgorithm should be cached to avoid performance overhead.

[gemini-code-assist]

There is a critical bug here due to the parameter renaming. The parameter was renamed from transportChannel to inputChannel, but transportChannel.isDirectPath() was not updated to inputChannel.isDirectPath().

As a result:

1. transportChannel now refers to the instance field this.transportChannel, which is null by default, leading to a NullPointerException when withTransportChannel is called.
2. Even if this.transportChannel is not null, it incorrectly uses the isDirectPath property of the old channel instead of the new inputChannel.

It should be updated to inputChannel.isDirectPath().

Suggested change

	transportChannel.isDirectPath(),
	inputChannel);
	inputChannel.isDirectPath(),
	inputChannel);

[gemini-code-assist]

Since System.nanoTime() can return negative values (as its origin is arbitrary), initializing lastRefreshTimeNanos to 0 can cause the first refresh (and all subsequent refreshes until System.nanoTime() becomes positive) to be skipped. This is because now - 0 will be negative, which is always less than TimeUnit.SECONDS.toNanos(5).

To fix this, you can treat 0 as a sentinel value indicating that no refresh has occurred yet.

Suggested change

	long now = System.nanoTime();
	if (now - lastRefreshTimeNanos < TimeUnit.SECONDS.toNanos(5)) {
	LOG.fine("Channel pool was refreshed recently, skipping duplicate refresh");
	return;
	}
	lastRefreshTimeNanos = now;
	long now = System.nanoTime();
	if (lastRefreshTimeNanos != 0 && now - lastRefreshTimeNanos < TimeUnit.SECONDS.toNanos(5)) {
	LOG.fine("Channel pool was refreshed recently, skipping duplicate refresh");
	return;
	}
	lastRefreshTimeNanos = now == 0 ? 1 : now;

[gemini-code-assist]

Querying environment variables via System.getenv() on every retry check can introduce significant performance overhead in the hot path. Since environment variables do not change during the application's execution, it is highly recommended to cache this value in a private static final boolean field.