WW-4817 Makes better decisions on methods first call

[yasserzamani]

if check permission and run synchronized on method first call in order to fix WW-4817

[lukaszlenart]

Tested locally and looks good, didn't notice any problems :)

[yasserzamani]

Thanks for your test. Yes, I think these changes do no have any functional test in point of view of outside of the invokeMethod method. Previously it got decision then cached, now it caches then decides :)

Just one more thing that I think maybe we do not have to call _securityManager.checkPermission(getPermission(method)); for each method call and the first one for caching is enough but I decided to keep it unchanged logically as I was not sure.

[lukaszlenart]

Please open PR against 3.0.x branch as well and then please open it again against the master branch and then you can improve the _securityManager.checkPermission(getPermission(method)); call :)

[yasserzamani]

Sure :) Do you mean that calling _securityManager.checkPermission(getPermission(method)); in each method call is not necessary (should be removed from 3.0.x, 3.1.x and master) and the first one (i.e. in method first call and filling cache) is enough?

[lukaszlenart]

No, I meant to keep changes as minimal as possible in 3.1.x and 3.0.x - so your current approach is fine. But in your third PR against the master branch you can optimize this call just to first method call - the master branch is still an experimental and a bit unstable and it targets 3.2.x series.