burst_width less than sample_rate / fft_size should be an error

[jamesoncollins]

Setting burst_width below sample_rate / fft_size will cause d_burst_width to be 0 and will later result in errors.

gr-fhss_utils/lib/fft_burst_tagger_impl.cc

Line 206 in 57c4b83

d_burst_width = burst_width / (sample_rate / fft_size);

[NZSmartie]

Been running into this issue myself, and can confirm that there needs to be a check for these parameters.

With d_burst_width resulting in 0, line 589 will cause a segfault:

gr-fhss_utils/lib/fft_burst_tagger_impl.cc

Lines 575 to 590 in 57c4b83

	size_t search_start =
	(size_t)std::max((int)factor * (b->center_bin - d_burst_width / 2), 1);
	// Ensure that search_stop is not 0 by performing ceil() division on
	// (d_burst_width / 2) note: int(true) == 1 in C++
	size_t search_stop = std::min(factor * (b->center_bin + (d_burst_width / 2) +
	int(d_burst_width % 2 != 0)),
	(size_t)(d_fine_fft_size - 2));
	// Normalize the magnitude by the noise floor
	for (size_t i = search_start; i < search_stop; i++) {
	d_fine_magnitude_shifted_f[i] /=
	d_baseline_sum_f[i / factor] / d_lookahead;
	}
	float* max_entry = std::max_element(d_fine_magnitude_shifted_f + search_start,
	d_fine_magnitude_shifted_f + search_stop);

search_stop will end up being 0, which is less than search_start and so std::max_element will crash