Include TICKLESS stack size increase even without LPTICKER_DELAY_TICKS

[LMESTM]

Description

LPTICKER_DELAY_TICKS extra stack size has been introduced with the LP
ticker C++ wrapper. Now we've removed the wrapper and LPTICKER_DELAY_TICKS
definition, but we still need extra stack size for the low level wrapper.

This change is needed for #10701

As such the change would impact many targets, more than expected, so maybe there is a better way to create the condition - feedback welcome

Pull request type

[x ] Fix
[ ] Refactor
[ ] Target update
[ ] Functionality change
[ ] Docs update
[ ] Test update
[ ] Breaking change

Reviewers

@mprse

Release Notes

[ciarmcom]

@LMESTM, thank you for your changes.
@mprse @ARMmbed/mbed-os-core @ARMmbed/mbed-os-maintainers please review.

[adbridge]

@mprse @ARMmbed/mbed-os-core could we have a review please ?

[bulislaw]

Maybe we could set idle-thread-stack-size-tickless-extra to be 0 by default and set other value for platforms that need it in targets.json?
@mprse do you know is ST platforms would be the only ones the old condition would be true right now? In any case should be easy to check.

[mprse]

This is a general code for all targets - should not be removed.

[LMESTM]

Hi @0xc0170 - I think we would benefit from your help and guidance to know which direction to take for this PR :-)

[0xc0170]

Where is this coming from? Looking at #10701 - I dont see there where this extra stack size is needed ? If delay ticks are set to 0, what else needs extra stack in the idle loop?

[LMESTM]

Where is this coming from? Looking at #10701 - I dont see there where this extra stack size is needed ? If delay ticks are set to 0, what else needs extra stack in the idle

The HAL LP ticker driver's code has grown in size because it does about the same checks that C++ Wrapper was doing, so now it uses most probably more stack and gets called from IDLE stack size....

[0xc0170]

Based on that, looks like the extension to the idle stack size should be provided for a target - rather tha removing this delay. something like @mprse proposed above: `if target sets extra idle stack, add it to the idle stack calculation).

[LMESTM]

PR updated with a new proposal - @mprse @0xc0170 something like that ?

[0xc0170]

Ci started

[mbed-ci]

Test run: SUCCESS

Summary: 11 of 11 test jobs passed
Build number : 1
Build artifacts