Calibration of ADC readings `HIT` to energy `E` works like this
```
E = MULI HIT * gain
E = ADDI HIT + pedl
```
The `pedl` was calculated as
```
pedl = -$floor(p*gain)
```
where `p` is the ADC reading corresponding to E = 0 keV.
Since `p` is a large magnitude negative number, the limited precision
of the `MULI` instruction parameter causes significant errors in `pedl`.
This patch aims to fis those errors. The new formula for `pedl` is
```
pedl = $floor(-p*$MRND(gain))
```
i.e., we use the same value for `gain` as the `MULI` instruction. This
is now in place for any automatic calibration calculations in
`step_temp_calib.l3`.
For the FS in space we use a fixed precomputed calibration. This is based on
the TVAC FS calibration for T=-36°C, with some adjustments for the changed
`acq_time=6`. The TVAC calib has been saved earlier to
`step_fs_calib-36.calib` with full precision, i.e., without taking `$MRND()`
into account. The awk script `step_fs_calib-36.awk` reverts those
calculations, applies the adjustments, and outputs the adjustes numbers. Since
awk does not have access to `$MRND()`, the script has been changed to output
formulas for `pedl`, including `$MRND()`,
