Setup Wizard: More helpful error state when trying to import from unprovisioned device

[nucleogenesis]

Summary

2024-07-02.16-07-59.mp4

• Takes string from here "No learning facilities found"
• Uses that by making the "Select facility" show an error when there are no facilities
• Adds hideContinue prop to OnboardingStepBase to hide the Continue button there when we have this error state
• Fixes issue where user's selection was lost when clicking "Go back" from the error state

References

Fixes #12066

Reviewer guidance

See the issue comment here for replication path -- note that I kinda had to go back and forth a few times through the flow - never actually importing anything - before I got the device to show up.

Overall, does the workflow feel okay?

[github-actions]

Build Artifacts

Asset type	Download link
PEX file	kolibri-.pex
Windows Installer (EXE)	kolibri-0.17.0a0.dev0+git.193.g31750c0d-windows-setup-unsigned.exe
Debian Package	kolibri_0.17.0a0.dev0+git.193.g31750c0d-0ubuntu1_all.deb
Mac Installer (DMG)	kolibri-0.17.0a0.dev0+git.193.g31750c0d-0.4.2.dmg
Android Package (APK)	kolibri-0.17.0a0.dev0+git.193.g31750c0d-0.1.3-debug.apk
TAR file	kolibri-0.17.0a0.dev0+git.193.g31750c0d.tar.gz
WHL file	kolibri-0.17.0a0.dev0+git.193.g31750c0d-py2.py3-none-any.whl

[pcenov]

Thanks @nucleogenesis - I confirm that now there's an error message if I attempt to import a learner from unprovisioned device.
Just have one question - in #12066 I have mentioned that the expected behavior would be for the device to not be shown in the 'Select device' modal which is the case for example when I attempt to import a full facility from that same unprovisioned device. Is that not an option in this case?

[bjester]

To @pcenov's note, we should probably add a prop to SelectDeviceForm for filtering locations that have a facility, i.e. filterHasFacility. Then have the useDeviceFacilityFilter apply the filter. That should prevent it from ever getting as far as where these error messages would show.

That would take care of filtering devices that would not be joinable or importable because they have no facility, which is a solid improvement. Although it doesn't go quite as far as the original issue's proposal of filtering those devices that actually have learners for import. I think the consensus was that we don't need to go that far. See my comment in regards to that