Fix DPP code to allow us to update to latest DPP

[pifopi]

Hello! I've been using DPP for a C++ project for a while now, and it worked great. Recently we updated from 10.0.22 (I think) to the 10.1.3 release and have experienced segfaults.

Repro steps:
Start the bot.
Stop the bot shortly after (before shards are ready).

Segfault happens in for (uint32_t s = 0; s < numshards; ++s) (cluster.cpp line 345). The crash does not happen if I let the shards fully initialize.

I create std::unique_ptrdpp::cluster m_bot = nullptr; before initializing/configuring/starting the bot. Stopping the bot calls shutdown() before resetting the pointer (can provide actual code if needed).

This wasn't an issue before, so seems like some sort of regression. Is there a way to check when shards are ready in order to prevent premature destruction? The on_ready() event triggers before the shards are ready.

copied from https://discord.com/channels/825407338755653642/825411707521728512/1426722220780163204

Small code example reproducing the issue (forked from their example)
https://github.com/pifopi/windows-bot-template

[pifopi]

#875 is helping to achieve this by allowing anyone to seamlessly update dpp (binaries and header and be able to compile against Pokemon Automation code

[pifopi]

brainboxdotcc/DPP@0d5e932
Is the commit that broke DPP (I reproduced a small example locally and I was able to find this)

[pifopi]

The issue is a this ptr is taken in the lambda and this can go out of scope while the lambada still capture it. Changing unique_ptr to shared_ptr leads to a huge leak though
pifopi/DPP@3327d14

Waiting for completion works so I opened two PRs and I hope one of them will be merged
brainboxdotcc/DPP#1522
brainboxdotcc/DPP#1523