Investigate EventCode implications

[apcarp]

I recently found a bug in myquery that stemmed from a possible misunderstanding of mya event codes. I discussed with Chris Slominski how to interpret event codes. Turns out the first and second nibbles of the event code are orthogonal. 0x00, 0x01, ... 0x05 are codes for why there is not data in this update (0x00 for there is data) and 0x00, 0x10, ..., 0x30 are for why there is not data prior to this point (0x00 for there is data prior to this point).

The main point is that any 0x*0 event code will have a valid data value associated with it. That's different than just 0x00 having a data point to use. This will probably impact a variety of methods and structures like streams and downstream dependencies.

myquery issue: JeffersonLab/myquery#11

[apcarp]

From the c++ source code, we have essentially two dimensions to a mya event. One is the reason why there is no data (or if there is data) and the other is the reason why there is previous data (or if there is previous data).

Codes for reasons for data/no data with related explanations.

ET_Data = 0x00;
ET_Disconnect = 0x01;
ET_Off = 0x02;
ET_Shutdown = 0x03;
ET_Unknown = 0x04;
ET_NaN = 0x05;
{
  "Normal channel data point",
  "Network disconnection",
  "Archiving of channel turned off",
  "Archiver shutdown",
  "Unknown unavailability",
  "NaN/infinity encountered"
}

Codes for reasons for no prior data/there is prior data with associated descriptions:

ET_NotFirst = 0x00;
ET_Inception = 0x10;
ET_Offline = 0x20;
ET_Discard = 0x30;
{
  "There are previous data values available",
  "Origin of channel's history",
  "Channel's prior data moved offline",
  "Channel's prior data discarded"
}

Also, there is a general mya "rule" that a table can only start with an update that has data. A sensible approach probably involves splitting these codes into two orthogonal enums in java and allowing downstream dependencies to check for the presence of data without having to have a list of which codes or nibbles indicate data presence.

Based on discussions with Chris Slominski, no message is presented for a 0x00 code even though two messages are defined as above.