r537: model the next base to GT/AG

[PMID:18688272] shows that the base following GT tends to be A or G (i.e. R) in
both human and yeast, and that the base preceeding AG tends to be C or T (i.e.
Y). In the new model, we pay no cost to GTr..yAG, but we pay half of the cost
if there is no r or y. This improves the junction accuracy when mapping to
human and mouse and decreases the accuacy when mapping to SIRV. My guess is
that SIRV does not honor this trend. Need to investigate in future.

Also in this commit, --cost-non-gt-ag is aliased to -C. The default is changed
to 9 instead of 5. I also added --splice-flank to enable the above model. This
may become the default once I confirm my hypothesis on SIRV.

minimap2.1

@@ -1,4 +1,4 @@
-.TH minimap2 1 "22 October 2017" "minimap2-2.2-dirty (r531)" "Bioinformatics tools"
+.TH minimap2 1 "27 October 2017" "minimap2-2.3-dirty (r537)" "Bioinformatics tools"
 .SH NAME
 .PP
 minimap2 - mapping and alignment between collections of DNA sequences
@@ -219,6 +219,9 @@ costs
 .RI min{ O1 + k * E1 , O2 + k * E2 }.
 In the splice mode, the second gap penalties are not used.
 .TP
+.BI -C \ INT
+Cost for a non-canonical GT-AG splicing [0]
+.TP
 .BI -z \ INT
 Break an alignment if the running score drops too quickly along the diagonal of
 the DP matrix (diagonal X-drop, or Z-drop) [400]. Increasing the value improves
@@ -239,9 +242,6 @@ both strands;
 .BR n :
 no attempt to match GT-AG [n]
 .TP
-.BI --cost-non-gt-ag \ INT
-Cost of non-canonical splicing sites [0].
-.TP
 .BI --end-bonus \ INT
 Score bonus when alignment extends to the end of the query sequence [10].
 .SS Input/output options
@@ -371,8 +371,8 @@ is that this preset is not using HPC minimizers.
 .B splice
 Long-read spliced alignment
 .RB ( -k15
-.B -w5 --splice -g2000 -G200k -A1 -B2 -O2,32 -E1,0 -z200 -ub --cost-non-gt-ag
-.BR 5 ).
+.B -w5 --splice -g2000 -G200k -A1 -B2 -O2,32 -E1,0 -C9 -z200
+.BR -ub ).
 In the splice mode, 1) long deletions are taken as introns and represented as
 the
 .RB ` N '