bug: Handle legacy simplepush records as candidate webpush records

[jrconlin]

Some older records in the database appear to have been miscategorized as
"simplepush". We should try to handle those as "webpush".

Closes #1033

[codecov-io]

Codecov Report

Merging #1034 into master will increase coverage by <.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master    #1034      +/-   ##
==========================================
+ Coverage   99.76%   99.76%   +<.01%     
==========================================
  Files          55       55              
  Lines        8799     8808       +9     
==========================================
+ Hits         8778     8787       +9     
  Misses         21       21

Impacted Files	Coverage Δ
autopush/tests/test_endpoint.py	100% <ø> (ø)	⬆️
autopush/web/registration.py	100% <100%> (ø)	⬆️
autopush/web/webpush.py	100% <100%> (ø)	⬆️
autopush/tests/test_integration.py	100% <100%> (ø)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update a9a1330...a0c7dfa. Read the comment docs.

[pjenvey]

Why not just reject these? Isn't that what previously occurred?

61b5752#diff-ccbbd52ff880cf4ddb71b18d126d189cL89

[bbangert]

I think we shouldn't reject them because the client thinks they're valid. We don't actually verify that router_type is correct in websocket.py. We probably should verify that and force a reset if the router_type is wrong.

[jrconlin]

These seem to be edge cases. I'm not exactly sure how they got their route_type set to simplepush since they have a /wpush url. Passing them on as a webpush is fairly harmless since any additional issues would be caught by the appropriate checks later.

Eventually, when we start enforcing the global TTLs we can drop this patch.