Cluster Specific Trampoline Data Structure

[Xeratec]

Overview

It would be nice to have a mapping between the hartid and the type of core. This can potentially be coupled with additional information about the cluster and supported features.

This will allow the optimization of the data structure used to offload functions to the Snitch cluster by only statically allocating them for the number of Snitch clusters in the design.

Discussion

FUTURE: Thinking of mixed-cluster or mixed-inter-cluster-ISA chimeras, we might want to have this explicitly only be for snitch cores.

Originally posted by @Scheremo in #13 (comment)

This would indeed be a clean solution, but I am wondering how to implement this. If this is specific to the Snitch clusters, we need a data structure (either at compile time or runtime) to map the hartid to the device (core) type. I would leave this to another PR and live with the unnecessary storage space for the non-Snitch clusters.

Originally posted by @Xeratec in #13 (comment)