Add CheckTx API

[tjanez]

Closes: #1555.

[kostko]

I think we should not handle checks as separate API methods. The reasoning is that execution also must run the exact same check before proceeding with execution which would lead to some duplication and could be easily forgotten by the runtime implementors.

So instead we should just add a flag (e.g., check_only) to the transaction Context and then format the methods as follows:

pub fn insert(/* ... */) -> Fallible<Option<FooBar>> {
    // ... perform checks here ...
    if ctx.check_only {
        return Ok(None);
    }
    // ... actual execution here ...
    Ok(Some(result))
}

The only downside is that this adds Options to all results as otherwise you need to generate a valid return value in that check only branch, so for the example insert call it would look like this (since the API call also uses an Option in its result type):

pub fn insert(args: &KeyValue, ctx: &mut TxnContext) -> Fallible<Option<Option<String>>> {
    // ...
}

An alternative could be defining a special error that would signify that the check was successful and this error would be transparently handled by the dispatcher and convert it into a successful result with no output value, for example:

use ekiden_runtime::transaction::dispatcher::CheckOnlySuccess;

pub fn insert(/* ... */) -> Fallible<FooBar> {
    // ... perform checks here ...
    if ctx.check_only {
        return Err(CheckOnlySuccess.into());
    }
    // ... actual execution here ...
    Ok(result)
}

I think I would prefer this last approach.

[tjanez]

An alternative could be defining a special error that would signify that the check was successful and this error would be transparently handled by the dispatcher and convert it into a successful result with no output value, for example:

use ekiden_runtime::transaction::dispatcher::CheckOnlySuccess;

pub fn insert(/* ... */) -> Fallible<FooBar> {
    // ... perform checks here ...
    if ctx.check_only {
        return Err(CheckOnlySuccess.into());
    }
    // ... actual execution here ...
    Ok(result)
}

I think I would prefer this last approach.

I've refactored my PR to follow this approach. @kostko, please take another look.

[codecov]

Codecov Report

Merging #1675 into master will not change coverage.
The diff coverage is 100%.

@@           Coverage Diff           @@
##           master    #1675   +/-   ##
=======================================
  Coverage   55.22%   55.22%           
=======================================
  Files         200      200           
  Lines       19082    19082           
=======================================
  Hits        10538    10538           
  Misses       7461     7461           
  Partials     1083     1083

Impacted Files	Coverage Δ
go/worker/common/host/protocol/types.go	20% <ø> (ø)	⬆️
go/worker/common/host/mock.go	85.36% <100%> (ø)	⬆️
go/worker/compute/committee/node.go	48.78% <100%> (ø)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 0335143...efc26fc. Read the comment docs.

[codecov]

Codecov Report

Merging #1675 into master will decrease coverage by 0.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master    #1675      +/-   ##
==========================================
- Coverage   56.46%   56.45%   -0.02%     
==========================================
  Files         206      206              
  Lines       19548    19548              
==========================================
- Hits        11038    11036       -2     
- Misses       7360     7361       +1     
- Partials     1150     1151       +1

Impacted Files	Coverage Δ
go/worker/common/host/protocol/types.go	20% <ø> (ø)	⬆️
go/common/runtime/runtime.go	100% <ø> (ø)	⬆️
go/worker/compute/committee/node.go	48.78% <100%> (ø)	⬆️
go/worker/common/host/mock.go	85.36% <100%> (ø)	⬆️
go/roothash/tests/tester.go	94.25% <0%> (-0.96%)	⬇️
go/storage/client/client.go	59% <0%> (ø)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update e31a1dc...5f3c681. Read the comment docs.

[kostko]

Nice work, this looks pretty good! Only some minor nits and after you fix those and mark the PR as ready for review it is good to merge!