Also allocate as many interms and outterms with helper functions as possible.

Now it is possible to add parameters multiple times and we no longer depend
on the location of the parameter in the list.

	modified:   controller/ec/esc_device.c

Whoops. Make sure 'str' is initialized before we try to print it.

	modified:   controller/controller/controller_sample.c

Use send_log(.., "%s", linebuf) instead of send_log(.., linebuf) to prevent
argument parsing since there is no way to pass arguments to the function
calls.

	modified:   controller/ec/block_stoeber.c

Initialize 'ret' to the most likely correct value (0). This at least makes the
code take the same path on every invocation. Let's just hope it's the correct
path :)

	modified:   controller/ec/esc_id.h

Eekk. Quickly close the lid on this can of worms.

	modified:   controller/ec/esc_id.h

Replace illegal character encoding of capital o-umlaut in Stoeber's name with
escape sequence. This silences a warning in clang.

	modified:   controller/ec/block_beckhoff_el4xxx.c

Add scale values for all other variants and add where the info came from. It
is evident that Beckhoff has published errors in this data, see the comments
in the code. That is why all these entries will remain 'unverified'.

	modified:   controller/ec/block_beckhoff_el4xxx.c

Give the driver a higher chance of actually detecting EL40[1/2]* blocks. The
product IDs were off.

	modified:   controller/ec/block_beckhoff_el4xxx.c

Fix table values for the EL400[1/2/4]. Since I only have the EL4004 to verify
the code, leave the EL400[1/2] 'unverified' for now. (The values are correct
according to Beckhoff documentation, but I can't test if these blocks also
work with the syncmanager change.)

	modified:   controller/ec/block_beckhoff_el4xxx.c

Oops. Fix missing argument.

	modified:   controller/ec/block_beckhoff_el4xxx.c

Make the initialization of the block actually work. The block offers two DC
configurations and at this moment, the ec code picks the second configuration
and probably forgets to select it in the right CoE register at this moment.
At this moment we asume all EL4xxx blocks need this fix. We could be wrong
about that. Leave all 'driver_verified' bits except for the EL4008 disabled'
for now.

	modified:   controller/ec/block_beckhoff_el4xxx.c

The device type table has a big number of off-by-one and off-by-a-lot errors.
Disable support for all blocks for now that haven't been explicitly tested and
verified to work correctly.

	modified:   controller/ec/block_beckhoff_el4xxx.c

Remove copy/pasto. The EL400* blocks were in the table twice.

	modified:   controller/ec/block_beckhoff_el4xxx.c

Fix the driver for the EL4008. The values we output to it should be limited to
32767, not 4096.

	modified:   controller/ec/esc_sm.c

Explicitly disable syncmanagers for empty PDOs.

	modified:   controller/block/build.mk

Add the 'pand' block.

	modified:   controller/controller/controller_sample.c

o Disable cpu_bind() on FreeBSD. The interface is different on FreeBSD. Leave
  implementing it for a future visitor to this file.
o Switch to CLOCK_MONOTONIC on FreeBSD. My test system doesn't implement
  CLOCK_MONOTONIC_PRECISE.
o DOn't set the stack size of new threads to PTHREAD_STACK_MIN on FreeBSD.
  Things will explode otherwise.

	modified:   controller/controller/build.mk

Don't try to link against 'libdl' on FreeBSD. dlopen() and friends are in libc
there.

	modified:   controller/block/build.mk

Don't try to link against 'libdl' on FreeBSD. dlopen() and friends are in libc
there.

	modified:   common/log/log.c
	modified:   common/utils/weather.c
	modified:   controller/controller/controller_command.c
	modified:   controller/controller/controller_trace.c
	modified:   controller/shell/shell.c

	Don't use PTHREAD_STACK_MIN as thread stack size on FreeBSD.
	PTHREAD_STACK_MIN must be huge on Linux for all this code to work. On
	FreeBSD the code will generally hit the end of the stack before the
	third function call.

Introduce first two position aliases
Allow string variables to be used almost anywhere

Don't double load js files.
Hopefully fix 100% cpu load bug in httpd.
Make sure changed type traces get a default value.

Use immediate trigger in test.ctrl files.
Fix some bugs

Cleanup libs
Add lib dependancies for console_httpd.
Add inf as double constant.

Some small fixes in lex code.

Lots of test updates

Fix some setpoint generator quirks

Move /dev/mem code to single file