Profile the in-for-loop mutexing in the main for-loop of point reading

[petrapa6]

Investigate and profile whether these mutexes in pub_handler.cpp are not slowing down the pipeline. If yes, optimize the code.

[veskrdan]

Feature implemented in commit f114b65 and improved speed 2x.

Notes:

• Mutexing in each for loop was indeed slowing the method down
• Solution was to store points in reserved vector first and then append it to the storing vector with .insert()
• No need for .reserve() here, since the final size can be calculated by the .insert() function

[petrapa6]

@veskrdan How big was the speed gain during your tests?

[veskrdan]

@petrapa6 About 0.010 ms on average. The method was about 0.020 ms, now circa 0.010 ms.