Fix issue with ESAVE after using FILL_KNOB

[tpersson]

This fixes the issue reported in #1149 from @fsoubelet. However, it would be good if someone else checked if this has some unwanted side effects.. I haven't noticed any but there is definitely a risk there is..

[rdemaria]

After discussing with @ldeniau:

current_node = NULL was added to fix another bug

The issue is that

augment_count

is used in

pro_error_make_efield_table

and if t->p_nodes is not null, then it relies on current_node to be meaningful.

[tpersson]

I think this is all good right? Should we merge?

[rdemaria]

ok for me!