node-rpc-ext: Remove rayon (Fix deadlock)

[kvinwang]

We have got a deadlock in the khala-node while getting storage changes.
According to the gdb stack frame dump, we found the deadlock is caused by the following flow:

[time 0] Thread A: locked(rayon default thread pool) in `pha_getStorageChanges`.
[time 1] Thread B: locked(sp wasm runtime cache) in block executor.
[time 2] Thread B: acquiring (rayon default thread pool) in `fn runtime_version`.
[time 3] Thread A: acquiring (sp wasm runtime cache) `fn runtime_version`.

Ideally, this can be solved by replacing the rayon thread pool used in pha_getStorageChanges with a separate pool rather than the default one.
However, due to the thread pool API of rayon:

thread_pool.install(|| {
        let result = headers
            .into_par_iter()
            .for_each(|| { /*do the work*/ })
            .collect();
})

The codes running in the inner closure inherit the installed thread_pool automatically.
So it would still deadlock in this single thread like:

[time 0] Thread A: locked(rayon custom thread pool) in `pha_getStorageChanges`.
[time 1] Thread A: locked(sp wasm runtime cache) in block executor.
[time 2] Thread A: acquiring (rayon custom thread pool) in `fn runtime_version`.  < deadlock here

Turns out this is an aged issue laying there for a long time.

I've also tried to replace the rayon with a different version:

rayon = { version = "1", git = "https://github.com/rayon-rs/rayon.git" }

But it failed to compile due to that they depend on the external lib.

So, let's just remove the rayon at the moment.

@h4x3rotab @jasl

[jasl]

LGTM

[h4x3rotab]

LGTM