fix: drop okhttp internal hostname verifier import, pin hostname verification, harden insurance test

[AndyScherzinger]

Two follow-ups from the okhttp 5 update (#6500) and the flaky test fix (#6927), plus a test pinning hostname verification.

1. Stop importing okhttp's internal OkHostnameVerifier

RestModule wrapped okhttp3.internal.tls.OkHostnameVerifier, an internal okhttp class without compatibility guarantees that a later release may move or remove. The same default verifier is now read through the public OkHttpClient.hostnameVerifier(). In okhttp 5.5.0 OkHostnameVerifier is a Kotlin object and the client's default (OkHttpClient.Builder.hostnameVerifier), so the very same instance is wrapped as before and hostname verification is unchanged, including the wrapping by our TrustManager for user-accepted certificates.

The new RestModuleHostnameVerificationTest builds the client through RestModule.provideHttpClient and runs real certificates (okhttp-tls HeldCertificate) through its hostname verifier:

• a certificate for the requested host is accepted
• a certificate for another host, for a subdomain, or with the host only as common name is rejected
• a certificate for another host is only accepted once the user trusted it via the certificate dialog (the existing TrustManager fallback)

Unlike TrustManagerTest, which mocks the default verifier, this fails if hostname verification is ever weakened, whether by an okhttp default or by the client setup. Checked by temporarily installing an accept-all verifier in RestModule: all four rejection tests fail.

2. Make the insurance request never() checks observe the IO thread

In OfflineFirstChatRepositoryTest, the insurance request reaches the network mock from ChatMessageSyncer's pull flow on Dispatchers.IO, which runCurrent() does not wait for. A never() check right after runCurrent() therefore passed even when a request had been started too early, as long as the IO thread had not reached the mock yet: the mirror image of the race fixed in #6927. The three checks now use after(LONG_POLLING_OBSERVATION_MILLIS).never(), so they watch for 200 ms before concluding that nothing was sent.

Verified by injecting a bug (insurance requests no longer skipped while paused, plus a 100 ms delay on the IO thread): the old test still passed, the new one fails as it should. Without the injected bug, the class passes repeatedly (about 0.6 s slower in total); ktlint and detekt are clean; gplay and generic compile.

🖼️ Screenshots

🏚️ Before	🏡 After
B	A

🚧 TODO

• Smoke test on a device: log in and open a chat on a server with a valid certificate and, if available, on one with a self-signed certificate you accept

🏁 Checklist

• ⛑️ Tests (unit and/or integration) are included or not needed
• 🔖 Capability is checked or not needed
• 🔙 Backport requests are created or not needed: /backport to stable-xx.x
• 📅 Milestone is set
• 🌸 PR title is meaningful (if it should be in the changelog: is it meaningful to users?)

🤖 AI (if applicable)

• The content of this PR was partly or fully generated using AI

[github-actions]

📱 QA build

Download	app-qa-debug.apk
QR code	Open the QR code for this download
Commit	1a0e9b1
Version	6928
Available until	7 days after this build

The QA build installs alongside a released Nextcloud app, so you can keep
using your existing install while testing.

Downloading the file requires a GitHub account, so open this link on the
device you want to test on, or transfer the APK to it.

[github-actions]

Codacy

Lint

Type	master	PR
Warnings	138	138
Errors	15	15

SpotBugs

Category	Base	New
Bad practice	7	7
Correctness	8	8
Dodgy code	40	40
Internationalization	3	3
Malicious code vulnerability	3	3
Performance	8	8
Security	11	11
Total	80	80