Before this change, class hierarchy is like this when using mixin:
```
class MyClass(BaseClass, Mixin1....)
```
which is usually ok when classes don't override each other's

But if we want to overwrite BaseClass's method by using mixin to do more thing,

For example:
```
class BaseClass(object):
  def test(self):
    print 'base test'

class Mixin1(object):
  def test(self):
    super(Mixin,self).test()
    print 'mixin'
```
I want to display 'mixin base test' when call test, but it doesn't work

since priority of how methods are resolved is from left

to right: BaseClass----->Mixin1, it only display 'base test'

So the correct way to use mixin should be in reverse order:
```
class MyClass(Mixin1, BaseClass)
```

/reviewed-on nexedi/erp5!935

When other nodes takes time to start they are not always registered
when the "main" node asserts that they are there.
We observed problems in an environment where resolving localhost (by
DNS) takes 10 seconds.

Also introduce getOtherZopeNodeList utility method that can be useful in
other ZEO tests in ProcessingNodeTestCase

When other zope nodes takes time to start this sometimes causes
conflict.

Reset mailhost before each test.

Commit changes created by portal_transforms.

This allows restoring the correct skin selection in ERP5JS.
Deferred reports can now be used in ERP5JS.

REQUEST is used to support direct form rendering, like: ./Base_viewHistory?deferred_portal_skin=ODS&portal_skin=Deferred

Thanks to Georgios Dagkakis for his work on this topic.
See !702

otherwise it loses too much precision.

Allow to sync opml event when some password are missing or broken (will only display public informations)

When adding opml from SlapOS interface, sync then show promise status.

Show Ressource consumption graph into separated graphs. Auto update of graph content is disabled by default, a new option is added to enable if needed.

Fix changes after merge with erp5 master which contains latests changes on renderjs

Intead of using opml software type, Hosting Subscription has now his own software type which simplify management.

[erp5_web_monitoring] increase opml import limit, show state of imported Instance and if they are monitored or not

[erp5_web_monitoring] Fix concurence problem on destroy OPML while sync is running or start just after

If Sync start in background and destroy OPML is launched, some data can be removed while sync is writing. So we prevent sync to run if destroy OPML is running and also, we refuse to destroy OPML if sync is running.

If the search result is only one item, redirect to the item view page, else redirect to the list of that items and let user choose the good one.

Monitor Jio storage has been updated to fix monitoring global Json files and set correct portal_type to it. The fix allow to keep backward compatibility with old/non updated monitoring backend instance.

This reverts commit ec926ea5.

This commit is still experimental, it was not supposed to be part of
this push.

This breaks matrix index consistency.
Also, call isBasePricePerSlice rather than introspecting wrapped method
arguments - the only wrapped method is _setBasePricePerSlice anyway !
Also, trigger SupplyCell_updateSliceBasePrice as it depends on
isBasePricePerSlice value on this line, and it just changed, and that
workflow does not trigger on line changes (which is a bug).
Also, get rid of a trivial local variable.

Rather than deleting them to just immediately create new ones with the
same ids but different uids.
Also, modernise code and follow naming conventions and coding style.

"self._m" may trigger "implicit self argument" magic. Follow how the
rest of this class accesses this property.