Move read lock after hasHeadState

[terencechain]

This fixes a bug raised in #7423. hasHeadState already calls the read lock so dead lock could happen. This moved read lock to after hasHeadState

[codecov]

Codecov Report

Merging #7427 into master will decrease coverage by 0.05%.
The diff coverage is 42.85%.

@@            Coverage Diff             @@
##           master    #7427      +/-   ##
==========================================
- Coverage   60.12%   60.07%   -0.06%     
==========================================
  Files         419      419              
  Lines       30835    30231     -604     
==========================================
- Hits        18540    18161     -379     
+ Misses       9228     9082     -146     
+ Partials     3067     2988      -79     

[farazdagi]

Another way to solve it:

• Define exported HasHeadState() which will lock the mutex (basically it will be what current non-exported hasHeadState() is).
• Define non-exported hasHeadState() it will be lock free version, used internally in package, where lock is already obtained.

That way, everywhere where you know that you've obtained the lock (and author of the package normally knows its contents well), you use lock-free version (if lock is already acquired). And from outside package, you are guaranteed that access is controlled by lock.

That way you don't have to worry about order of statements (first check head, then obtain locks as in the current PR), as you will simply use lock-free version, if lock is already obtained.

The only downside of that method, is exposing HasHeadState() (not too much of an issue, but still).

[farazdagi]

• In order to avoid duplication, can we use lock-free version of func in exported one (I've added corresponding suggestion)
• Removed bunch of added blank lines (if you've added them intentionally, please, ignore suggested code changes)