refactor(crypto): Simplify the verification object constructors

[poljar]

This patch moves all the important identities that we need for verification into the already existing IdentitiesBeingVerified struct.

It adds a method to get those identities to the store, and streamlines our to-device and in-room verification constructors.

This is the first part of #399 and incidentally fixes a bug where the started_from_request flag wouldn't be set correctly for in-room verifications. The buggy constructor has been removed.

[codecov-commenter]

Codecov Report

Merging #632 (2a5f170) into main (b2273ab) will increase coverage by 0.29%.
The diff coverage is 87.75%.

@@            Coverage Diff             @@
##             main     #632      +/-   ##
==========================================
+ Coverage   74.18%   74.47%   +0.29%     
==========================================
  Files         101      101              
  Lines       15369    15539     +170     
==========================================
+ Hits        11401    11573     +172     
+ Misses       3968     3966       -2     

Impacted Files	Coverage Δ
crates/matrix-sdk-crypto/src/identities/device.rs	73.36% <ø> (ø)
crates/matrix-sdk-crypto/src/identities/user.rs	66.00% <ø> (+0.99%)	⬆️
...atrix-sdk-crypto/src/verification/sas/inner_sas.rs	49.22% <ø> (+1.35%)	⬆️
...tes/matrix-sdk-crypto/src/verification/requests.rs	54.30% <78.94%> (-2.70%)	⬇️
...rates/matrix-sdk-crypto/src/verification/qrcode.rs	70.94% <79.16%> (+0.31%)	⬆️
crates/matrix-sdk-crypto/src/verification/mod.rs	64.84% <90.90%> (+1.17%)	⬆️
...ates/matrix-sdk-crypto/src/verification/machine.rs	76.47% <100.00%> (-1.61%)	⬇️
...ates/matrix-sdk-crypto/src/verification/sas/mod.rs	80.09% <100.00%> (-0.76%)	⬇️
...atrix-sdk-crypto/src/verification/sas/sas_state.rs	87.32% <100.00%> (-0.07%)	⬇️
crates/matrix-sdk/src/error.rs	17.85% <0.00%> (-1.38%)	⬇️
... and 16 more

---

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 b2273ab...2a5f170. Read the comment docs.

[jplatte]

Might also make sense to turn the boolean into a dedicated enum, but those calls have definitely become more readable than before this PR, so no need to block it on that.

[poljar]

Might also make sense to turn the boolean into a dedicated enum, but those calls have definitely become more readable than before this PR, so no need to block it on that.

Sure, I'm not closing #399 yet. There's plenty of more cleanup this part of the code should receive.