bug: confusing error logs for non-top VM frames

[itegulov]

CallErrorTracer prints errors for all VM frames even if they do not make the transaction fail.

Example:

08:04:46  INFO Validating 0xc2d617ef17ed5ebb0fd59255891bf9a638decf27b37811c2e8a15cd360f3b14a
08:04:46  INFO Executing 0xc2d617ef17ed5ebb0fd59255891bf9a638decf27b37811c2e8a15cd360f3b14a
08:04:46 ERROR !! Got error flags: 
08:04:46 ERROR WRITE IN STATIC
08:04:46  INFO 
08:04:46  INFO ✅  [SUCCESS] Hash: 0xc2d617ef17ed5ebb0fd59255891bf9a638decf27b37811c2e8a15cd360f3b14a

Where WRITE IN STATIC happens inside a simulated static call somewhere deep inside the VM (non-top frame)

[dutterbutter]

Related to: #436

I'd like to refactor CallErrorTracer to return result with accumulated errors something like below. We only care about the returned flag to use to provide a more informative error message. For example, if ERROR NOT ENOUGH ERGS flag is returned we know the user ran out of gas during execution and can more explicitly inform them of what action to take.

pub struct CallErrorTracer {
    result: Arc<OnceCell<ErrorFlags>>,
}

impl CallErrorTracer {
    pub fn new(result: Arc<OnceCell<ErrorFlags>>) -> Self {
        Self { result }
    }
}

impl<S, H: HistoryMode> DynTracer<S, SimpleMemory<H>> for CallErrorTracer {
    fn after_decoding(
        &mut self,
        _state: VmLocalStateData<'_>,
        data: AfterDecodingData,
        _memory: &SimpleMemory<H>,
    ) {
        if !data.error_flags_accumulated.is_empty() {
            let _ = self.result.set(data.error_flags_accumulated);
        }
    }
}

impl<S: WriteStorage, H: HistoryMode> VmTracer<S, H> for CallErrorTracer {}