core: reduce monomorphization of the futures code

[yukibtc]

These changes will reduce the monomorphization of the futures code, resulting in a reduction of the binary size.

In my case, the binaries for android targets have become 100/200kb smaller.

[bendk]

Thanks for the patch it's a really interesting idea. I think we should try to think through the trade-offs here a bit more before landing this one.

I hope to put together a bunchmark on async call overhead. This is my main question: how much overhead does boxing the futures add per-call? The other half of the trade-off is the binary size, can you share some details about your project so we can understand that better? Like, how many async functions are you exporting?

Once we know both of those, I think we'll be in a much better position to judge this one. Maybe the boxing overhead will be small enough that we can just hit the merge button. Maybe it's a real trade-off and we should add a config option for this.

[yukibtc]

I hope to put together a bunchmark on async call overhead. This is my main question: how much overhead does boxing the futures add per-call?

I'll try to add some benchmark soon.

The other half of the trade-off is the binary size, can you share some details about your project so we can understand that better? Like, how many async functions are you exporting?

This is the project: https://github.com/rust-nostr/nostr/tree/d030b776f521fc2bb032f9490bab89e3bfbb5108/bindings/nostr-sdk-ffi

I tried to count the async fn with a script and the result is 117 functions.

This is the result of cargo bloat --release -n 1000 of my project compiled for x86_64-unknown-linux-gnu target using uniffi b46ab54:

cargo-bloat-uniffi-b46ab54.txt

While this is the result using the patch (uniffi a6c2f67):

cargo-bloat-uniffi-a6c2f67.txt

[yukibtc]

I'll try to add some benchmark soon.

I tried to take a look at the current benches but I'm not sure how the new ones should be added

[bendk]

Thanks for those details, it's very interesting to go through. I'll try to put together an async benchmark in the next week or two.