Exposure computation wrong for GBM data >~ 2017

[giacomov]

evtbin/src/MultiSpec.cxx

Lines 147 to 159 in bcd2493

	KeyValuePairCont_t::iterator found2 = m_key_value_pairs.find("EVT_DEAD");
	KeyValuePairCont_t::iterator found3 = m_key_value_pairs.find("EVTDEDHI");
	if (m_key_value_pairs.end() != found2 && !found2->second.empty()) {
	double deadtime;
	found2->second.getValue(deadtime);
	double evtdedhi;
	found3->second.getValue(evtdedhi);
	double gbm_exposure=(gti.computeOntime())-(total_counts*deadtime)-(total_error_channel*evtdedhi);
	(*table_itor)["EXPOSURE"].set(gbm_exposure);
	}
	total_counts=0;
	total_error_channel=0;
	}

I discovered this bug investigating a request to the Fermi helpdesk (number 3694).

Circa 2018 the keyword EVT_DEAD and EVTDEDHI went respectively from 2.6E-6 and 1.04170E-5 to 2.6D-6 and 1.04170D-5, i.e., from single precision to double precision floating point.

This trips gtbin into computing the exposure wrong.

This happens in the lines linked above for PHA2 spectra, but I think it is more general about the way keywords are read (probably assuming single precision floating point).

[donhorner]

@dagorym Can you take a look at this one?