Limit input length for decode command

[SkymanOne]

Closes #980

The hex bytes are consumed for decoding values of arguments on demand. If there is not enough data to fill the buffer, the error is already thrown. Previously, spare bytes were just ignored. Since this PR, an error will be thrown if input buffer contains spare bytes.

Here is the small demo demonstrating that extra bytes in buffer do not affect decoding of value.

//boolean is 1 byte long
let value = true;
let encoded: Vec<u8> = vec![1, 12, 3];
let mut encoded: &[u8] = encoded.as_ref();
let decoded: bool = bool::decode::<&[u8]>(&mut encoded).unwrap();
// is true
assert_eq!(decoded, value);
//prints [12, 3]
println!("{encoded:?}")

[ascjones]

The alternative approach would have been to do the bytes length check just once directly in decode.rs after the decoding is complete. This way is also okay, with adding it to each of the called methods, although as mentioned decode_contract_event is used in one other place so we need to check whether that still works.

[SkymanOne]

Extracted check to a separate function instead of moving it to a higher layer because each decode function returns a generic anyhow error. We would need to return a custom error and map all anyhow errors inside each decode function to a specific error. We will also need to handle the error on the higher decode function. Additionally, events and messages/constructors have different structs that are used to extract label. That will also unnecessary overcomplicate error handling.

Therefore, it is better that all validation happens on the lower level at each decode function independently.

[ascjones]

LGTM

[ascjones]

Since @HCastano had also reviewed this, would also have been nice to ask for his approval too.