Pub sub plugins

See merge request cds/advligorts!91

Update a comment to correct a name.  Put a magic number into a constant.

1) For now, the RCG will accept either ifo= or site=, eg ifo=X2 or site=X2 to designate the IFO in the cdsParameter block. This
    obviates the need for immediate changes to present Matlab control models. All new models, and eventually all old models,
    should use the ifo= designator as the site= designator will be deprecated in a TBD future release.
2) At compile time, if a model still contains the site= designator, a warning message will be written to alert user to change to ifo=.
     This is simply a warning for now.
3) I have gone thru all the RCG Perl code to change site to ifo, where appropriate.  I think things are consistent now ie site is a
     location eg lho, llo, lao, etc. and ifo is the IFO designator eg ifo=H1, L1, A1, etc.

This is the daqd memory format, daq_multi_cycle_data_t in daq_core.h.

Cds_pub_sub now understands input from rmipc:// (FE mbuf inputs) and output to daqm:// (daqd memory buffers, daq_multi_cycle_t in a mbuf).

Created cds_pub_sub from cps_recv
Added a sub plugin for the rmIpcStr based mbufs from the FE systems.

- Added check in ADC initialization code.
- Passed back as a new ADC status bit (3)
- Added status bit to GDS_TP screen ADC status
- Since difficult to remember what bits mean, ADC status bit info
  added with labels to the auto generated MONITOR_ADC screens.
- Note that this bit is only reported by IOP models.

systemd: set CAP_SYS_NICE for awgtpman service

Closes #103

See merge request cds/advligorts!89

closes #103

fix capitalization in systemd UMask setting

See merge request cds/advligorts!90

provides a listing of available Parameter block settings.

IOP: setting virtualiop=1 will force all IPCs defined in the model to
	become SHMEM type regardless of model definintion.
USER APP: setting use_shm_ipc=1 will force all IPCs defined in the model to
        become SHMEM type regardless of model definintion.
See README file in advligorts/src/fe directory for fulldetails of
Parameter block options.

Close Issue #87 - delete setup sub-folder, already moved to new cds/rts-admin project

See merge request cds/advligorts!86

Fix a typo in installing mbuf_probe, put it in bin not BIN

See merge request cds/advligorts!85

   - Deprecated old, non-functional on Debian 10, controllerVirtual.c.
   - Parameter block entry "virtualIOP=1" sets the appropriate compile flags.
   - controllerIOP.c modified with addition of RUN_WO_IO_MODULES option
   - Added timing control in time_slave_io.c
Further changes are still required to feed data into the ADC inputs.  As is,
ADC inputs are statically set to value of channel number eg channel 31= 31.

Closes #98.  Check for dcuids that are not normal and abort.

Closes #98

See merge request cds/advligorts!82

Changing cmake to install mbuf_probe so that it can be put into packages.

See merge request cds/advligorts!83

mbuf_probe is a useful debug tool that is typically built from source, but could be useful in any FE/daqd system to examine issues.

This check can be turned off by setting the parameter "allow_any_dcu" = "1" in the daqdrc.

This allows dcuid's in the channel list that are not in the TP group or the MYRICOM group.  (We may be able to restrict the TP group as well).

Also removed some old code that is specific to the internal edcu, which along with not being used anymore gives a more obscure error message.

allow awgtpman to run as unprivileged user

Closes #93

See merge request cds/advligorts!81

The awgtpman process is run as root only because it wants to bind to a
privileged port.  This is not necessary as capabilities exist to allow
running it as a unprivileged user:

* remove an unnecessary check in the awgtpman binary on the superuser
  status of the user

* set CAP_NET_BIND_SERVICE in the systemd service to allow it to bind
  to privileged port

This should allow the service to run as a non-privileged user, which we
set here to be advligorts.

closes #93

support: set umask 0002 for rtcds and epics service

Closes #90

See merge request cds/advligorts!80

This should make sure that all files created by these commands (e.g. in
/opt/rtcds/...) are created group writable permission.

closes #90

I reenabled def for this macro in src/include/inlineMath.h.
The only model that seems to use this is ASC, which would not compile without
this change.

rtcds: export all variables loaded from /etc/advligorts/env

See merge request cds/advligorts!79

This wraps the sourcing of the /etc/advligorts/env file in the 'allexport'
shell option that explicitly marks all variables for export.  Hopefully
this will eliminate the need to explicitly export variables (address #92).

awgtpman depends on the module being loaded, so a Wants seems like the
appropriate strength.

  - Moved send_io_info_to_mbuf routine from map.c to moduleLoadCommon.c, as the
    TIME_SLAVE option does not use map.c.
  - Changed controllerIop.c to not write DAC bit registers if TIME_SLAVE, as
    this area is not mapped for this configuration ie no I/O modules.

New dolphin packages install symvers to a particular directory.  This change...

See merge request cds/advligorts!78

New dolphin packages install symvers to a particular directory.  This change looks for those symvers and incorporates them into model builds.  Note that configure was changed by autoconf.  Configure should probably be dropped from git.