[Pijukatel]

Summary

Syncs the .NET client with OpenAPI spec v2-2026-11-04T151257Z (apify/apify-docs#3083, #3084, #3152) and apify-client-js #1179, #1187, #1188, #1190 and #1197. The client version goes to 0.9.0 (breaking: PaginationList.Count has a new meaning).

Spec sync

• The embedded schemas are regenerated. The only response-schema change is actorPath on Actor versions and builds.
• New model properties: ActorVersion.ActorPath, Actor.StandbyUrl, Actor.ActorStandby, Actor.TaggedBuilds and Build.ActorDefinition.

Server Actors (JS #1179)

• ActorClient.Server() returns an ActorServerClient with RequestAsync, GetUrlAsync and GetOpenApiSpecAsync.
• New types: ActorServerRequestOptions, ActorServerResponse and ActorServerException.
• The token goes in X-Apify-Authorization.
• Every response is returned, including error statuses.
• Only 429 and 503 are retried. Network errors and timeouts are not.
• Redirects are not followed.
• The JS client's deprecated standby() alias is not ported, because the .NET client never had a Standby() method.

Redirect handling (new public API)

• ApifyRequestOptions.FollowRedirects is set on requests that must not follow redirects.
• HttpClientTransport accepts an optional second HttpClient, with redirects disabled, for those requests.
• A transport that wraps your own HttpClient without that second client throws InvalidOperationException for such a request. This is documented.

Page Count (JS #1187, #1190), breaking

This matches the reference client and the Rust, Go, Java and PHP clients.

• On a dataset items page, PaginationList.Count is the larger of X-Apify-Pagination-Count (rows scanned) and the number of items returned.
• On every other list, Count is the count the API reports, as in the JS, Java and Go clients. It equals the number of items, except on the Store listing, where it can be larger.
• Offset iteration steps by Count, capped at the requested limit. Dataset iteration stops when Count is 0, as in JS.
• IReadOnlyCollection<T>.Count, LINQ Count(), enumeration and the indexer still count only the items returned. Code that indexes a page up to Count must use Items.Count instead. The CHANGELOG and docs say this.
• The iteration docs are updated (JS #1190).

Signed URL validity (JS #1188)

• A zero or negative expiresInSecs for signed public URLs now throws before any request is sent.

CI

• The CI and publish workflows run the unit tests with TZ=Asia/Tokyo, and fail when a test filter matches no tests.