Insert test for too few data to avoid ungracefull crashes

[kmk]

With GRAS Raw sampling data it was observed that certain profiles would come out of the merge CL + RS with very few data points which would cause crashes later (especially when there's only 1 data point)

Suggested solution is to insert af test for <100 data points (following OCC). This does potentially reject marginally useful data (but even at 50 hz 100 points is just 2s of data)

I suggest inserting the following lines into ropp_pp_preprocess at line 147 (immediately upon commencing section 5)

---

IF (ro_data%Lev1a%npoints .lt. 100) then

CALL message(msg_warn, "Error: Too few data") config%obs_ok = .FALSE. RETURN

ENDIF

---

Potentially the test could be for an even smaller number of points (e.g. 10) I don't think anything more than maybe 2 or 3 would actually cause a crash.

[Ian Culverwell]

We'll go with Kjartan's first suggestion of needing at least 100 lev1a points in the profile. Easy to reduce later if necessary.

(As always, we're very grateful to Kjartan for suggesting a solution as well as a problem!)

It doesn't come into effect in our standard tests.

Implemented in ROPP5.1.