There are 64 channel per parameter, 7 parameters per VMM, 2 VMMs per hybrid, up to
6 hybrids per FEN, up to 32 FENs per ring and up to 12 rings per RMM.

We can't scan this without saturating the network card interface, so i removed it.
It's now up to the user to force the read. We do have forwawrd links between the setpoint and
readback records, however.

After talking more with the firmware developers, it became clear that we
can actually read from the registers when VMM is acquiring, we can only
not Set other parameters.

I still leave this as a commit instead of rebasing/amending to make it
easily reversible if we need it someday.

Now most pertinent RBV records are not processed when acquisition is
ongoing.
A few records like LinkStatus and IsAcquiring are still scanned for
obvious reasons, but if we find out that this could not be the case, we
change it.

I still think this should also be treated in the API layer.

ADCVal and AnalogMon should process in the correct order for us to know
what we are reading.

We use PHAS field to make sure everything is processed after we have
checked that IsAcquiring is 0.

Scannig was not added until now because RMMAPI had an issue that made it
scramble answers from the hardware if too much polling was made. This
issue was fixed so we add scan now.

In the channels array records, this called for a new record creation.
This is because we need the asyn:READBACK info tag, which makes the
readback from hardware and the setpoint record have the same values. So
I guarantee that the record ending with R is the actual scanned
parameter. I also removed the forward link from S to RB record to
guarantee that the latest value we have in $(CH)-S record is still the
setpoint.

AnalogMon and ADCVal-R have PHAS 0 and 1 respectively to guarantee that
we know what is the meaning of what we are reading.

Add polling to ADC value

See merge request !8

For now I avoid polling too many records because of a behavior described
in ticket ICSHWI-19975.

However, this is an important parameter, so I decided to poll it.

Major refactor

See merge request !7

In order for VMM channel records to be able to see
"#CalcDISA" record, they need to know the prefix for Hybrid records.
This new naming structure allows that.

asyn READBACK tag allows for waveform record to be updated whenever the
callback is called in code.

Without this, OPIs usually can't properly put or get to uninitialized
waveform records.

For user to be able to visualize what values are converted into
parameters.

Added enable/disable option, which now requires for all records to be
instantiated so we can properly communicate.

This particular IOC spits some error messages with the current test
setup because it only has two hybrids.

In this case the semicolon ended up being part of the property name in
the db file. Of course, this is not good but I wanted to avoid having to
add $(C)=':' for now.

Its easy to change in the future if needed.

Added logging details to several functions. Also added log messages to
IOCMessage parameter.

As noted on st.cmd, instead of naming records <Device>:HybN we will name
them <Device><N>:<Property> (So the number is actually part of the
device).

So records with two digits refer to the whole IOC, records with three
digits refer to hybrid-specific parameters and records with four digits
refer to VMM3A specific parameters.

It was looking for parameter VMM_FEN_ACQUIRE but not finding it.
Now it can properly read the parameter, although it shouldn't be needed
to be read most of the times...

All parameters that share the same ADC are now added: Pulser, Threshold,
Temperature and Bandgap.

All except temperature are showing weird values though, so that needs to
be investigated. I suspect something to do with configHybrid or configFE
or configVmm, but I am not sure.

This reads Temperature from API memory and sets the structure for future
parameters such as BandgapADC, etc...

The timestamp is also read and put into a PV in order to be explicit for
user when was the last time it was read from hardware.

Now if API returns a bad status, asyn knows about it.

To use the correct term.

removed proof of concept. Made it work for all VMMs of all hybrids.

Turns out it wasn't that different from others. All it needs is to
change a bo record to ao.

Didn't add SD because the mechanism is a bit different.

This commit was done with an inescrupulous "git add ." after several
radical modifications.

Anyway, the API was radically modified and basically the old way the
code worked does not make any sense anymore.

The desired architecture is the following: the API is going to be used
as a class component. Each parameter will be created according to the
desired number of enabled hybrids.

Each parameter should use one simple API function. No information about
the VMM or hybrids itself is stored in the VMMTbl class unless extremely
needed.

This commit leaves a few old parameters that should be removed in the
future such as ADC_VALUE_VMM1, READ_ADC_VMMS, etc. This is only to serve
as a reminder to add those later.