fix undriven X-IF signals

[davideschiavone]

No description provided.

[cairo-caplan]

Thank you for the PR, @davideschiavone: looks good to me!

Now, regarding the modification of the X_HARTID_WIDTH used for the width of hart_id, which is used by the eXtension InterFace signals:

1. This change affected the eXtension InterFace IO ports, that causes the current SEC script to fail, since these ports are present on cve2_top, even if they are not used.
2. I don't think it is ideal to set it on the cve2_pkg, as a system integrator may want to use custom widths for the hartid, also considering different CVE2 instantiations, but leaving it at the maximum allowed width (of MXLEN) is fine. One can still exclude the extra bits on the ports association outside the core.

No changes requested on my side, but I would like to:

1. First inform @FrancescoDeMalde-synthara of the change, as a user of the CVE2 with the CV-X-IF.
2. Inform you that I created PR #305 , to update the SEC script to be able to check the design both with the eXtension InterFace disabled, which it passes, and with it enabled, which it does not pass due to the width changes of the IO ports.