Fixes for HWPE subsystem

[LuigiGhionda]

This PR introduces some modifications/fixes to the HWPE subsystem.

In particular:

1. It fixes the Astral-based pulp cluster in the following non-default scenarios:

   • N_HWPE = 1
   • N_HWPE = 0 (HwpePresent=0)
   • ECCInterco = 0

2. It introduces softex version that supports ECC-HCI

In addition, it also removes a duplicated call to vopt in the Makefile

[LuigiGhionda]

@ricted98 Since you redesigned the HWPE subsystem for Astral maybe you have some comments/suggestions

[ricted98]

Hi @LuigiGhionda , two questions:

• What was broken in these scenarios?
• Does the static mux not work in the "degenerate" case of a single output?

[LuigiGhionda]

@ricted98

* What was broken in these scenarios?

The problem is with hwpe_sel_int signal definiton which fails when N_HWPE=1 (https://github.com/pulp-platform/pulp_cluster/pull/88/files#diff-6bfef46bbce2076251409cb7cccedbf816d10bcf9dbd66fdb552501a62fd7837L69)

* Does the static mux not work in the "degenerate" case of a single output?

Yes, it can handle the degenerate case, but since its select signal is hwpe_sel_int, I thought to separate the two cases in order only to define it for N_HWPE!=1. However, that part could be modified

[ricted98]

So the problem is the +: operator? If so, we could also think of just adding a parameter to use the mux in all cases.
HWPE_SEL_BITS = N_HWPE > 1 ? $clog2(N_HWPES) : 1. What do you think?

[belanoa]

If the problem is just the +: operator, I would go with the solution suggested by @ricted98 . By the way @LuigiGhionda , is there a plan to merge your changes to HCI so that we can start cleaning up the dependencies?

[LuigiGhionda]

Yes, so I will follow your suggestion. Thank you!
I will take care of it as soon as possible