tests: segment_simul.fa may have a double D

Back in april 2013, when @mikael-s designed this simulated sequence,
he premonitorily included a borderline case of a VDDJ recombination.
There are here 6-7 nucleotides exactly conserved for the second D.

algo/tests/segment_simul.should_get

@@ -8,9 +8,14 @@ $ First sequence, easy segmentation (no error, few deletions at the windows, sma
 #              ||||||||||||                     ||||||||||||||||||||||||||||||||||||||||||||||
 # V5-10-1: 280 attactgtgcgaga  |||||||||||||
 # D1-14*01:  1              ggtataaccggaaccac
+#                                          |||||||
+# D3-16*01:          gtattatgattacgtttgggggaGTTATGcttatacc
 # J5*01   :  1                             acaactggttcgactcctggggccaaggaaccctggtcaccgtctcctcag
 # -2/CTTC/-3  -1/GTTA/5
 1:^>seq1 \\+ VDJ 0 72 77 89 94 139 IGHV5-10-1.0[1-4] 2/CTTC/3 IGHD1-14.01 1/GTTA/5 IGHJ5.01
 
+# Note that a second D (D3-16*01) can be detected (6 common nucleotides, or even 7 with an overlap on the first D)
+# This is tested in segment_simul.should-vdj
+
 $ seq1_polyT is the same as seq1 but starts and ends with 10Ts, the recombination is in the middle. start position should not be 0 and end position should not be length-1
 f1:^>seq1_polyT \\+ VDJ 10 82 87 99 104 149 IGHV5-10-1.0[1-4] 2/CTTC/3 IGHD1-14.01 1/GTTA/5 IGHJ5.01

algo/tests/should-vdj-tests/segment_simul.should-vdj.fa

 
 
 ### Copied from data/segment_simul.fa and algo/tests/segment_simul.should_get
+# See algo/tests/segment_simul.should_get for a detailed description
+# With only one D, this is:
+# IGHV5-10-1*01 2/CTTC/3 IGHD1-14*01 1/GTTA/5 IGHJ5*01  [IGH]
 
->IGHV5-10-1*01 2/CTTC/3 IGHD1-14*01 1/GTTA/5 IGHJ5*01  [IGH]
+>IGHV5-10-1*01 2/CTTC/3 IGHD1-14*01 1//23 IGHD3-16*01 8//7 IGHJ5*01  [IGH]
 aagtccatcagcactgcctacctgcagtggagcagcctgaaggcctcggacaccgccatgtattactgtgcga
 CTTC
 ataaccggaacca
 GTTA
-tggttcgactcctggggccaaggaaccctggtcaccgtctcctcag
+TG
+gttcgactcctggggccaaggaaccctggtcaccgtctcctcag