[dlopezaldama]

Title: Fixes #377: GROUPR partial fission in residual production, sqrt(E) smoothing conditions, storage checks

Base branch: develop (rebased on NJOY2016.80, 51b1661).

What is fixed

1. Heavy-residual production (MF26/36) with partial fission (groupr issues #377). Fission was excluded from the
   residual-production lists only for MT18; MT19, 20, 21 and 38 are now excluded as well: in the input check
   (a message is printed and the request skipped), in conver and in the a>4 production list.
2. sqrt(E) extension of the low-energy neutron spectra (ismooth, MF6 LAW=1). The smoothing is applied only
   under the conditions that ACER (acelf6) already uses in 2016.80: no discrete lines and first outgoing energy
   equal to zero; lin-lin tables also need more than 3 points. Before, GROUPR checked only LEP and ZAP, so
   discrete lines were smoothed and the first point of spectra not starting at zero was overwritten; GROUPR and
   ACER are now consistent.
3. Storage. getsed checks the size of the tmp storage; nwscr 10000 → 20000, ncmax 1000 → 7000,
   ntmp 10000 → 20000; scr is deallocated at the end of groupr and conver.

Tests

• Rebased on develop (2016.80, 51b1661); replaces the earlier 2016.78-based commit of this branch (which
  conflicted with the 2016.79/.80 GROUPR changes).
• LANL test suite: 87/87 pass with the existing reference tapes (no tape changes).

Release notes entry

Proposed text for the next release section of ReleaseNotes.md (not added to the file in this branch, to avoid conflicts between the open PRs):

• GROUPR now excludes all partial fission reactions (MT19, 20, 21, 38), not only MT18, from the heavy-residual production (MF26/36) (issue groupr issues #377).
• The ismooth option of GROUPR now uses the same conditions as ACER: spectra with discrete lines or whose first outgoing energy is not zero are not smoothed.
• GROUPR array sizes were increased and storage checks added in getsed.