Unify client/server and NVM setup/teardown

[padelsbach]

First pass unification of repetitious test code. Net reduction is about 600 lines.

To ensure that this does not significantly alter the test flows, I generated a coverage comparison script. The output below shows effectively the same level of coverage in the unit tests between this branch and main.

➜ /git/experiments/hsm-coverage/compare_coverage.sh /git/wolfHSM/coverage_out/main_summary.txt /git/wolfHSM/coverage_out/padelsbach_client-server-test-unify_summary.txt
==============================================================================
  Coverage Comparison
  Baseline: main
  Target:   padelsbach_client-server-test-unify
==============================================================================

--- Files with changed coverage ---

  File                                       Coverage        Delta    Exec           Lines
  -----------------------------------------------------------------------------------------------
  /git/wolfHSM/src/wh_server_crypto.c         83% ->  83%  (+0%)  exec: 2184->2181  lines: 2630->2627
  /git/wolfHSM/src/wh_server_keystore.c       84% ->  84%  (+0%)  exec: 792->792  lines: 938->936

--- Summary totals ---

                         Baseline     Target      Delta
  ----------------------------------------------------
  Total lines               14255      14250         -5
  Lines executed            11964      11961         -3
  Line coverage               83%        83%        +0%

  Files in baseline: 43
  Files in target:   43

[bigbrett]

womp womp no malloc allowed :(

[bigbrett]

I unfortunately never informed you about the critical requirement that we don't malloc inside wolfHSM .

So while this would be wonderful to do, we can't :/

Gonna need to skin the cat another way. That said, pretty sure we don't need variably sized buffers for anything in our test harness - could we get away with default macro configuration structs?