Generalize RP2040 (and in the future, STM32) support to NUM_CORES

[sberkun]

Currently, the rp2040 support assumes that there is one main core and one other core:

reactor-c/low_level_platform/impl/src/lf_rp2040_support.c

Lines 223 to 258 in b58a3a5

	#define MAGIC_THREAD1_ID 314159
	void core1_entry() {
	thread_1_return = thread_1(thread_1_args);
	sem_reset(&thread_1_done, 1);
	}
	int lf_thread_create(lf_thread_t* thread, void* (*lf_thread)(void*), void* arguments) {
	// make sure this fn is only called once
	if (num_create_threads_called != 0) {
	return 1;
	}
	thread_1 = lf_thread;
	thread_1_args = arguments;
	num_create_threads_called += 1;
	sem_init(&thread_1_done, 0, 1);
	multicore_launch_core1(core1_entry);
	*thread = MAGIC_THREAD1_ID;
	return 0;
	}
	int lf_thread_join(lf_thread_t thread, void** thread_return) {
	if (thread != MAGIC_THREAD1_ID) {
	return 1;
	}
	sem_acquire_blocking(&thread_1_done);
	// release in case join is called again
	if (!sem_release(&thread_1_done)) {
	// shouldn't be possible; lf_thread_join is only called from main thread
	return 1;
	}
	if (thread_return) {
	*thread_return = thread_1_return;
	}
	return 0;
	}

However, future versions of this chip (i.e. an RP4040) may have more cores. Ideally, lf_thread_create and lf_thread_join should be revised to account for this. The generalization probably won't be perfect (since multicore_launch_core1 doesn't lend itself well to more cores), but should be aimed at a potential future with a 4-core RP4040. Something like the following may be used to abstract away multicore_launch_core1:

// assume this function will exist in some future version of the pico-sdk
void multicore_launch(int core, void (*entry)(void)) {
    multicore_launch_core1(entry); // for now, assume core == 1
}

This work on generalization will probably lend itself well to future STM32 support, as many STM32 devices do have more than 2 cores.