Failures in a deployment can lead to develop.element.io getting broken

### Steps to reproduce
Follow up ticket for https://github.com/vector-im/element-web/issues/26698

The build and deploy workflow failed halfway through (https://github.com/vector-im/element-web/actions/runs/7099975755/job/19325093703 I think) but left develop.element.io in a state where half the app was deployed. The effect was that where the files for the workers are normally returned directly (eg. for me right now, https://develop.element.io/bundles/696a45aec41256f38a68/blurhash.worker.js returns a 200) it was 301ing somewhere else (for me, https://develop.element.io/bundles/059b7536c78a459f9bf6/blurhash.worker.js is 301ing).

This appears to cause element to fail to load with an infinite spinner (see also https://github.com/vector-im/element-web/issues/26701).

Our CI deploy should not be able to leave the deployment broken (unless it's actually deploying a broken build of course): it should either deploy the new version or fail and leave the previous build deployed. It's possible this is cloudflare/pages-action not being very atomic?

### Outcome

#### What did you expect?

#### What happened instead?

### Operating system

_No response_

### Browser information

_No response_

### URL for webapp

_No response_

### Application version

_No response_

### Homeserver

_No response_

### Will you send logs?

No