Completes #6048: let FunctionalInterrupt.h support only IRAM-safe scheduled ISRs

[dok-net]

Makes for a much cleaner cores/esp8266/core_esp8266_wiring_digital.cpp:

• no special "functional" predicate
• no reverse "duplicate from functionalInterrupt.h keep in sync" copied-code dependency

[hreintke]

@dok-net @d-a-v @devyte
disclaimer : I did not read all the remarks from the last days as I was on a short break.

Doesn't solve #5922 the issue that std::function<void(void)> cannot be used in attachInterrupt due to not being in IRAM ?

[devyte]

it should. I've only taken a quick glance at the recent PRs, we're currently trying to get 2.5.1 out the door.

[dok-net]

@devyte @hreintke Maybe it's just me, I don't understand the point of using functional if all that's supported are static functions. Please run, commenting / uncommenting the various cases, the attached MCVE. I think it's a bug, misleading users into attaching ISRs that are not in IRAM, the whole point of #5995 is voided, too.
AttachInterruptArg.ino.txt

[d-a-v]

I don't understand the point of using functional if all that's supported are static functions

I agree.

You have lots of running PRs. Are they all related ? If so can you group them in one single PR ?
To be honest I'm quite lost :]

[hreintke]

I tried to get std::function::target running with rtti option set to verify myself but I only got crashes.

If I have a construct like

#include "InterruptTest.h"
#include "Arduino.h"
#include "FunctionalInterrupt.h"
#include "Schedule.h"

class InterruptTest
{
public:
	InterruptTest(int pin) : _pin(pin)
	{
		Serial.begin(115200);
		attachInterrupt(pin,std::bind(&InterruptTest::isr_iram,this),FALLING);
	}

	void ICACHE_RAM_ATTR isr_iram()
	{
		schedule_function([this]() { Serial.printf("isr_iram from pin %d\r\n",_pin); });
	}

	volatile int _pin;
};

InterruptTest D5Handler(D5);
InterruptTest D4Handler(D4);

void setup()
{
}

void loop()
{
}

Would that be a valid use (together with #5922 ?)

[dok-net]

@d-a-v This PR is more readable now (turn on diff settings: hide whitespace changes).
It is best reviewed as diff against https://github.com/dok-net/Arduino/tree/bugfix/attachInterrupt , or after PR #6048 was merged.

It should be less confusing now:

Bugfix first (#6048 ), then this PR.

OR:

As an alternative, that depends on the ICACHE_RAM dilemma surrounding the use of C++ functional solved in PR #5922, PR #6047 supersedes both #6048 AND this PR.

[dok-net]

Development of this branch stopped.
Cause: implementation based on std::function is more advanced, previous concerns involving IRAM-proofness have been addressed. Please go on with review/merge of #6047