tests: benchmarks: latency_measure: timing measurement values are all 0

[ABOSTM]

Describe the bug
Running tests/benchmarks/latency_measure on nucleo_f746zg,
test is passed, but all timing measurement values are 0 ns.

To Reproduce
Steps to reproduce the behavior:

1. west build -p auto -b nucleo_f746zg tests/benchmarks/latency_measure
2. west flash
3. See log

Expected behavior
Non null timing measurement value

Logs and console output

*** Booting Zephyr OS build zephyr-v2.4.0-1849-gc0650217cc7c  ***
START - Time Measurement
Timing results: Clock frequency: 72 MHz
Average thread context switch using yield                   :       0 cycles ,        0 ns
Average context switch time between threads (coop)          :       0 cycles ,        0 ns
Switch from ISR back to interrupted thread                  :       0 cycles ,        0 ns
Time from ISR to executing a different thread               :       0 cycles ,        0 ns
Time to create a thread (without start)                     :       0 cycles ,        0 ns
Time to start a thread                                      :       0 cycles ,        0 ns
Time to suspend a thread                                    :       0 cycles ,        0 ns
Time to resume a thread                                     :       0 cycles ,        0 ns
Time to abort a thread (not running)                        :       0 cycles ,        0 ns
Average semaphore signal time                               :       0 cycles ,        0 ns
Average semaphore test time                                 :       0 cycles ,        0 ns
Semaphore take time (context switch)                        :       0 cycles ,        0 ns
Semaphore give time (context switch)                        :       0 cycles ,        0 ns
Average time to lock a mutex                                :       0 cycles ,        0 ns
Average time to unlock a mutex                              :       0 cycles ,        0 ns
===================================================================
PROJECT EXECUTION SUCCESSFUL

Environment (please complete the following information):

• OS: Windows
• Toolchain gnuarmemb
• Commit SHA 1b65fd0

Additional context
Add any other context about the problem here.

[ABOSTM]

Issue is linked to DWT which never starts.
Writing to DWT registers is inefficient.
On Cortex M7 access to DWT register may be lock.
See #30327

[nashif]

@ioannisg FYI

[ioannisg]

Will take a look, sure