change stdout message print enables:
 fudgeflag2 enables info messages, fudgeflag3 enables error messages

fix sources_next list: sources_next index 1 is not a vaild value

added an entry for invalid sources_curr

changed max valid source_curr number

add ZED-F9T offset information to TIMELS comment

This driver uses the host's serial port RTS line to trigger the
module's EXTINT0 input. No kernel driver or PPS support is needed.

The module must be preconfigured as follows:
  UART to 9600 bps
  UBX-NAV-TIMELS transmitted once per second
  UBX-TIM-TM2 enabled
  CFG-TP5's gridUtcGps flag set for UTC

Performance is very good: On a lightly-loaded machine with (-24)
precision, 24-hour mean refclock jitter of 35 nanoseconds has been
 observed using ntpvis.

With this patch it is possible to trigger the receiver's event
input line with a parallel port controlled by a modified pps_parport
Linux kernel driver.  Latency should be reduced significantly relative
to using the normal serial port RTS output.
Jitter should also improve on highly-loaded machines, however
jitter is already low (less than 100ns using a Praecis Cf) on fast
lightly-loaded machines without the kernel driver.

the kernel driver can be found at
 https://github.com/retroman/pps_parport_polled

See the "Experimental PPS Feature" section in docs/driver_trimble.txt
for usage information.

This driver uses the host's serial port RTS line to trigger the
module's EXTINT0 input. No kernel driver or PPS support is needed.

The module must be preconfigured as follows:
  UART to 9600 bps
  UBX-NAV-TIMELS transmitted once per second
  UBX-TIM-TM2 enabled
  CFG-TP5's gridUtcGps flag set for UTC

Performance is very good: On a lightly-loaded machine with (-24)
precision, 24-hour mean refclock jitter of 35 nanoseconds has been
 observed using ntpvis.

See https://github.com/ThrowTheSwitch/Unity/releases for release notes

  It's not available on Fedora 31.
  I can't find where it is needed.

  If we find a case, we need to document it better.

  Limit to 25 total per restart.  (old was buggy)
  Dump whole packet.

It took me a while to figure out how the "phone" bit worked, even with
the documentation.  An example would have saved me time.

The documentation did not match the code: backup and auto were reversed.

Also, in the "Driver Options" section, "mode" was listed as "Not used by
this driver", which is incorrect.

At some point in the past, NIST supposedly dropped the on-time marker.
A comment was added as was support for a second format without the
on-time marker.  That was partially removeed in 2011 in commit
ef0eee67, but vestiges remained.

Signed-off-by: Mark Atwood <mark.atwood@ntpsec.org>

There was no error message if ntpd couldn't open the cert file.

A few printed (null) rather than a hostname when the
name was numeric.

Fixes NTPsec/ntpsec#620

It used to exit so the operator would know there was a problem
rather than having to debug things after noticing it wasn't working.
But exit-ing from a non-main thread crashes lib_getbuf()
which adds a layer of confusion.

Also repair/tweak a couple of log messages.

Provoked by Netnod's server announcement.
I guess that option doesn't get used much.