Add simulations & wagmi hooks

[Rubilmax]

Requests are faster because each of them is now independent (no need to fetch the config before fetching the rest of each market/vault). But deployless reads are not multicall-able, meaning they can't be batched in a single onchain RPC call via multicall.

This state is arguable and does not seems better out-of-the-box. However, here is why it is better:

• requests can be batched with most RPC backends nowadays
• requests are already actually batched using HTTP/2 natively via the browser (so there's no actual "delay" due to HTTP overhead)

So the only matter is the request count for RPC backends, which will be the cost paid by the RPC user in the end. It should stay relatively low considering that now, each entity costs at most 1 RPC request (either through the multicall = 0, or through a single deployless read = 1)

Fixes INTER-1016

[julien-devatom]

can't we let the user choose the method to use? either with multicall or with deployless

[julien-devatom]

Since we already know that deployless fails for mainnet weth, wdyt about always skipping deployless with a set of address? It can prevent known rpc fails

Else everything looks good

[julien-devatom]

I don't have reviewed logic yet but few remarks from your Thursday presentation:

blue-sdk-viem-simulation should not depend on react imo. Ex if we want to use it in liquidation SDK. A solution would be to have a dedicated package for the hook useSimulationState (which is ok I think)

• I think it's not necessary to have "viem" in the SDK name for a non primitive SDK like blue-sdk. underthehood, as soon as you want to use simulation, you use viem.