fix(widget): update local Widget Core pointer when changing core

[Kribylet]

Widget carried a Core* that wasn't being updated when changing Core, this commit fixes that. Widget::addFriend would add duplicate friends when restarting a Core, so a check was added to prevent adding duplicates of friends.

Fixes #5758

• Commits follow our git commit guidelines

---

This change is 

[sudden6]

Reviewed 1 of 1 files at r1.
Reviewable status: complete! 1 of 1 LGTMs obtained

[hart0]

I tried these changes and it fixes the graphical issues showing duplicated friends and stops the crashing when reconnecting, however attempting to send any message to a friend after the reconnect is done and my status is back online causes a crash.

Thread 1 "qtox" received signal SIGSEGV, Segmentation fault.
0x00000000006d327c in std::__invoke_impl<bool, bool (ICoreFriendMessageSender::* const&)(unsigned int, QString const&, NamedType<unsigned int, ReceiptNumTag, Orderable>&), ICoreFriendMessageSender&, unsigned int&, QString const&, NamedType<unsigned int, ReceiptNumTag, Orderable>&> (__f=
    @0x7fffffffcf60: &virtual table offset 24, __t=...,
    __args#0=@0x7fffffffcf5c: 0, __args#1=..., __args#2=...)
    at /usr/include/c++/9/bits/invoke.h:66
66      { return (__invfwd<_Tp>(__t).*__f)(std::forward<_Args>(__args)...); }

bt

[sudden6]

@Kribylet can this be closed now that we agreed on #5778?

[Kribylet]

@Kribylet can this be closed now that we agreed on #5778?
I believe so, since there would be no way to change the core of an application instance in that case.

[anthonybilinski]

Now that v1.17-dev is branched from master and #5778 has been merged to it, I think to continue to work towards (working) reconnect functionality in the future, backing out #5778 on master and continuing this work on top would be reasonable, targeting master branch with the PR.

v1.17-dev will have no reconnect button, but master branch and future releases can have a working reconnect button when this is finished.

[sudden6]

@anthonybilinski I'm against re-introducing the reconnect button for now, it's still not very useful.

[Kribylet]

Pushed the latest version of this PR from my local repository for potential use later, if we want to restore reconnect button functionality. This commit sequence changes the tox instance used by the Core instance, rather than replace the Core instance entirely.

I advocate harsh testing before applying this PR, but naively I wouldn't expect any issues with this that couldn't already occur from having a bad connection.

Personally tested using qtox normally, then reconnecting, resuming chat, and using AV features post reconnect.

[anthonybilinski]

@sudden6 even if it's fully working in a future release? I thought removing it was a temporary measure?

[jessica181920]

@sudden6 Clicking a button is much easier than closing the client and signing back in manually.