and make it the single object returned after a decryption

…ent id

ex: to_device events

and ignore others. Do not consider them as errors

MXCrypto.handleRoomKeyEvent() has been added to know exactly when we can start decrypting.

…nt data

It solves 2 things:
 - avoid synchronous decryption
 - decryption of the  MXRoomSummary.lastMessageEvent when it is edited (element-hq/element-ios/issues/4322)

Plus it saves a decryption.

to avoid useless (synchronous) decryption like in the previous commit

…hread

This will save synchronous decryption

…en done before

to avoid any synchronous decryption.

The algo is probably simpler.

It also fixes testGetLastMessageFromPagination, testGetLastMessageFromSeveralPaginations and testFixRoomsSummariesLastMessage tests that have been broken since b4d5ba7

…on update

lastMessageEvent is required by room list. It took 1.3 seconds on the main thread to get it on my account (600 rooms). Now there are asynchronously preloaded and decrypted when the SDK load.

One test has been fixed. As events are now decrypted only upstream and no more on demand, we need to trigger a decrypt to mimic this upstream decryption.

as it is not available in the decrypted JSON content

that does not pollute existing listeners on the live timeline

There can be only one component that paginates message from a time else there is will be races.

Use MXSession.decryptEvents to get decrypted events.

…o decryptEvent:inTimeline:]

…ith the decryption queue.

A single decryption takes 5ms on an iPhoneX. It was useless to synchronise this method with the busy decryption thread. There is no relationship.

Note that this operation uses 0.5ms. It must not be called from the main thread.
MatrixKit uses this method to build the timeline. It uses on a dedicated queue.

We must not call complete() before returning.
It breaks pagination in MatrixKit that uses the returned MXHTTPOperation.

This method is called from the main thread. It should not use directly elf.crypto.deviceList that is managed by the cryptoQueue.

Co-authored-by: SBiOSoftWhare <SBiOSoftWhare@users.noreply.github.com>

Co-authored-by: SBiOSoftWhare <SBiOSoftWhare@users.noreply.github.com>

End of the removal of decryption on the main thread