My advice would be to push to have them use an XML parser, a custom one that doesn't support the spec is just wrong and fudging results to comply just propagates bad practice even, as in this case, it can be done. There's little excuse nowadays for consuming XML using string parsing techniques.
And the two results are different:
is not the same as
as the first WORKORDER element is in a namespace whereas the second isn't. This means that using exclude-result-prefixes shouldn't work as it won't remove an actively used namespace declaration.
I will have to experiment, I've never used the #default or #all values.