With this MR, the gFex ByteStream decoder tool is updated to allow for multiple TOB slices (corresponding to consecutive BC). It previously worked in such a way that only one TOB slice per accepted event was decoded, with a fixed total number of gFex raw words decoded per event. This approach was leading to wrong decoding every time that the data frame contains multiple (typically 3) TOB slices. This is often the case for cosmics and calibration data. Now, depending on the slice number, the decoder is able to decode all kinds of data.
Tagging @rbielski to inform you that the zero words masking introduced in !56819 (merged) and discussed in ATR-25495 has been removed, as I realized that the only zero words that are now accepted and used to fill the containers are objects (Rho and JwoJ Scalar TOBs) that are not yet evaluated in firmware, and these objects being zero shouldn't throw any exception. To verify it, I tested the HLTSeeding code with the update, and no exceptions are indeed thrown. However, should this cause any problem, I will reintroduce (temporarily) the masking.