Fixing a small bug introduced in #200

[giovannipizzi]

When lcount=.true., some content of the
input file that was read was changed, and
the second run (e.g. if 'random' was present)
was giving wrong results.
Not clear because the error was errand, though
(was working on travis but failing on my machine)

[codecov]

Codecov Report

Merging #215 into develop will not change coverage.
The diff coverage is 33.33%.

@@           Coverage Diff            @@
##           develop     #215   +/-   ##
========================================
  Coverage    61.28%   61.28%           
========================================
  Files           29       29           
  Lines        17098    17098           
========================================
  Hits         10479    10479           
  Misses        6619     6619

Impacted Files	Coverage Δ
src/parameters.F90	81.04% <33.33%> (ø)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 16a82bb...d326b8f. Read the comment docs.

[VVitale]

Seems to work for me, not sure I understand the nature of the bug tbh

[giovannipizzi]

The input file is stored in a list of strings (and slightly modified to remove comments, fix capitalisation etc.)
While before the param_get_projections was called only once, so it wasn't a problem if it was modifying the list of strings, now it's called twice (once to count and once to get them) so it shouldn't change the input array.