Wrap-around problem

[walterhackin]

@MengRao,

The current wrap-around mechanism in the SPSC queue is implemented by resetting the write index to zero when there isn’t sufficient space at the end of the buffer—even though space might be available at the beginning. However, the approach of using a single contiguous static array forces the allocation of a very large continuous block of memory to achieve high capacity. This design not only wastes memory but also may be inefficient in managing the wrap-around, especially under high log generation rates.

Suggested Improvements:
1. Implement a segmented or ring-of-buffers approach to allow more flexible wrap-around without the need for a huge continuous memory block.
2. Optimize the logic to better recycle freed segments when the consumer advances the read pointer, ensuring that wrap-around is efficient and minimizes wasted space.

Sometimes it feels like poll() is clearing SPSC queue too long, so wrap-around just rewrite existing logs

[walterhackin]

And also i know that you can use LOGGER_BLOCK 1 to not lose data, but it is too complicated operation, and with huge amount of logs blocking make your program as slow as defauls ofstream works