Some issues in denoising and clustering

[tongtongcao]

Some issues in denoising and clustering were found:

1. SNR is still implemented for denoising, independent of AI-denoising. As discussed and tested, it should be cancelled.
2. Total number of raw DC hits is required to be less than 2000 for an event in raw data. It is dangerous for high luminosity. The limit should be limit for # of DC hits after AI-denoising.
3. In splitter for complicated hit clumps, it is required for candidate clusters that there is no skipped layer. The limit should be cancelled.
4. Prob. of linear fitting for clusters is required to be larger than 0.65. As tuning, the limit is not appropriate for 3-layer clusters.

[raffaelladevita]

addressed #494