Always insert instantiation statement at the end of the local statement list

[qobilidop]

This fixes #3955. Many test outputs need to be updated after this change.

[qobilidop]

I used P4TEST_REPLACE=True make check to update all test outputs. I'm still in the process of manually reading and checking these changes.

[qobilidop]

Ahh, I just see #821 that does the opposite change.

[mihaibudiu]

I used P4TEST_REPLACE=True make check to update all test outputs. I'm still in the process of manually reading and checking these changes.

The gauntlet tool does some of that, by I am not sure whether it will check such early passes.
(It performs translation validation, making sure that the program at two specific program points has the same semantics.)

[fruffy]

The gauntlet tool does some of that, by I am not sure whether it will check such early passes.

Gauntlet checks all passes of the front and mid end of p4test. That also includes type checking etc. So the inlining pass is checked. We actually have an issue open for that #2470.

The only thing it relies on is the parser. An unfortunate side effect of this design is that Gauntlet is hard to maintain 😞.

[qobilidop]

Just wonder if this PR needs more reviews or is it ready to be merged? I would appreciate it a lot if it could be merged soon, because I have some P4 code at work pending on this to compile. Thanks a lot for all reviewers' time!