python promises can define sense, test and anomaly method.
test method will be called by slapgrid or when no bang is needed for the promise anomaly method is
called when a promise failure required to bang the master, anomaly method can be optional but it's not the case for test method.

Slapgrid always run promises and save the result in .slapgrid/promise/result in a JSON format. The result will be used later by monitor
When a partition is correctly deployed, slapgrid will only run promise anomaly and will bang if there is an error and if the failed promise can bang

check promise anomaly when partition is upto date

This reverts commit b5a1e352

foo.com really exists and example.com was intended for this kind of
example (even if I'm not sure it's OK to actually connect)

Because this test really make a connection and the purpose is to test
URL generation, use a link local address that will fail immediately
rather than an unreachable address that will timeout.

To execute a slapconsole script, we could only do:
`slapos console < script.py`
but this does not stop in case of error

or
 `echo 'execfile("script.py")' | slapos console`
but this is too complicated.

Extend this API so that we can simply use

`slapos console script.py`

inspired by `python script.py`

use profile to log promises resource consumption

/reviewed-on !37

needed for example for wendelin.core arrays + connections

@alain.takoudjou , @rafael  if no objections can you please accept & merge it?

/reviewed-on !36

slapos/slap/slap.py slap.initializeConnection: Cache master node's Hateoas URL and do not ask many times. Communication to master node is expensive.

slapos/slap/slap.py: Fix ComputerPartition.setConnectionDict. Declare connection_parameter_hash explicitly, UnboundLocalError may occur.

rename checkPromises to checkPromiseList, and move the method to grid/utils so it can be imported by monitor to check promises.


`raise_on_failure` will be used to check all promises without raise. All promises result will be returned.

`profile` True/False will log promise execution time.

/reviewed-on !35

  The API uses code which is part of the new VIFIB 2.0 API, and
  uses simpler scripts and JSON for the exchange of information.

  This API may change again in futher future, but backward compatibility will be kept with the current implementaion

We only had a global "pidfile" configuration, but it's more conveniant
to be able to configure different pidfile for each command (software,
instance, report)

The supported config file would look like:

```
[slapos]
pidfile_software = /path/to/pid
pidfile_instance = /path/to/pid
pidfile_report = /path/to/pid

```



/reviewed-on !29

  Include a test for fetch configuration

  Conflicts on namespace can easily break this feature, so pick the
  template directly is safer.

This pluging can be used a add a script which wipe data before the partition is removed
The script run after all partition services are stopped. The script is ran by supervisord
then slapos node report will wait until it finish before destroy the partition.

/reviewed-on !23

This pluging can be used a add a script which wipe data before the partition is removed
The script run after all partition services are stopped. The script is ran by supervisord
then slapos node report will wait until it finish before destroy the partition.

/reviewed-on !22

Move manager.instance to the part of code which doesn't get preempted 

/reviewed-on !21

  This version include initial plugin system and cgroups integration.

  We also included a significant change for provide the partitions network informations, which now
  don't require to come from slapos master anymore, so it is possible to make local extensions.