Bump wasmtime default table_elements

[ggwpez]

Some Polkadot tests started failing (paritytech/polkadot#5788) with:

Failed to get runtime version: cannot create module: table index 0 has a minimum element size of 2093 which exceeds the limit of 2048

Investigation showed that the --feature=runtime-benchmarks flag causes the increase of the WASM size.
Can I just bump this value or do the others need an increase as well?
The tests work with the increase to 3072.

[shawntabrizi]

What other consequences does changing this have?

[koute]

What other consequences does changing this have?

Slightly more memory usage. But it shouldn't really matter as it shouldn't be a huge amount. (If push comes to shove we can always just extract the exact value necessary from the WASM blob and use that as a limit, since we instantiate a separate wasmtime::Engine per runtime anyway.)

[ggwpez]

bot merge

[bkchr]

@koute why not use the default number of table elements which is 10k? I have seen other teams having issues because of this number being to small.

[koute]

@bkchr Originally I didn't want to overcomplicate the code with autodetecting the necessary value (since it'd require some extra changes as the precompiled codepath used by PVFs doesn't currently have direct access to the WASM blob so it can't extract the value) so I decided to just pick a number and hardcode it for now.

As far as the number itself is concerned I wanted to be conservative since the impact of this is AFAIK technically multiplicative (the number of table elements times the configured maximum number of instances times how many runtimes we keep ready for instantiation). So I looked at what we need in production and just went with a double of that number.

If anyone's having issues with this limit please don't hesitate to ping me; for sure we can consider just increasing it if it's a problem.

[bkchr]

Yeah, I told them to open an issue, but they don't have done it. However, it should now be enough for them.