Optimize data-mutation-free for mutation observer

[medzhidov]

Optimization from this issue:
#2205

[neSpecc]

Please, implement a test for this case

[bettysteger]

@medzhidov can you write a test for this case? The test could be similar to this one.

[bettysteger]

@medzhidov so this will also resolve PR #2317 ?

[medzhidov]

@bettysteger Yes, but the solution suggested in #2317 is just another approach. I propose making the data mutation more user-friendly, while in #2317, it is suggested to create a new property for the block instance that will disable any mutations inside it. I believe both solutions are good, but the second one (#2317) is more fundamental because we can manipulate the block from its class instance. "data-mutation-free" is more versatile; it may be needed when we want to "hide" certain parts of our block from the onchange callback, rather than the entire block.

[bettysteger]

@medzhidov thanks for your answer! much appreciated!

I just did the "same" PR, but added a test: #2548

[neSpecc]

Resolved by #2548