Update surface to generic paved if footway and cycleway parts are different but paved

[mnalis]

This improves #4526 (originally closed by PR #4548):

• in situation when cycleway and footway are differently paved it will also update surface = paved (previously, it would just always delete surface tag in such case).
• actually removes smoothness when surface changes (we did test for it, but never actually did it)
• some more fixes for compilation error for tests that were present in vanilla v49.0 (introduced when different branches were merged)

(This is recommended way to tag different but paved surfaces, see second block in Ref= S5 on https://wiki.openstreetmap.org/wiki/Bicycle#Miscellaneous)

---

Compiles OK and manually tested that it works fine. It also includes automated test that pass too.

[westnordost]

Hey @mnalis, no need to request a review from me, I check the pull requests regularly. Marking as ready for review is enough. There are other regular contributors which may review the code as well.
If you request a review specifically from me, you make it look like only I would be able to review this code.

[mnalis]

Oh sorry, you're the only one the github shows by default, so it seemed inviting button to press 😃
I now see I can press the cog icon to request others too. Everyone is welcome to review of course, but if clicking that button is not required for processing of PR to proceed, than I won't be clicking it in the future. ❤️

(That request-review functionality is then probably less useful in opensource projects, and more useful in corporate environment with strict hierarchy when someone in power is selecting who should do what, I guess? At least I as lowly contributor would feel uncomfortable requesting by name someone more experienced to do the work, unless they previously explicitly expressed interest in that PR)

[westnordost]

Hm, I notice that I refactored the SurfaceAnswer.applyTo function to also remove smoothness if the surface was changed. Hence, part of this PR is obsolete/wrong.

However, I kept you waiting long enough and to save on further back and forth, I will make the changes myself before merging.

[smichel17]

@mnalis On review requests: I've found them useful in $dayjob not because of hierarchy but more like a shorthand for "I know you worked on this feature in the past, or have some stake in maintaining it, can you check if this is an ok change?" I think a similar thing applies here; someone might want to request my review when touching the recent answers code, for example.