distToNearest could use more better error messages

[Jason Vander Heiden]

See comments for details. The error message states invalid characters, when that is not the problem.

Original problem:
Works fine on linux, but in Windows 7 with R 3.2.4:

> library(shazam)
Loading required package: alakazam
Loading required package: ggplot2
> db <- subset(InfluenzaDb, CPRIMER %in% c("IGHA","IGHM") & 
+              BARCODE %in% c("RL016","RL018","RL019","RL021"))
> 
> # Use genotyped V assignments, HS1F model, and normalize by junction length
> dist_hs1f <- distToNearest(db, vCallColumn="V_CALL_GENOTYPED", 
+                            model="hs1f", first=FALSE, normalize="length")
There were 50 or more warnings (use warnings() to see the first 50)
> warnings()
Warning messages:
1: In value[[3L]](cond) : Invalid character in sequence. Cannot compute distance.
2: In value[[3L]](cond) : Invalid character in sequence. Cannot compute distance.
3: In value[[3L]](cond) : Invalid character in sequence. Cannot compute distance.
4: In value[[3L]](cond) : Invalid character in sequence. Cannot compute distance.
5: In value[[3L]](cond) : Invalid character in sequence. Cannot compute distance.
6: In value[[3L]](cond) : Invalid character in sequence. Cannot compute distance.
7: In value[[3L]](cond) : Invalid character in sequence. Cannot compute distance.
8: In value[[3L]](cond) : Invalid character in sequence. Cannot compute distance.
9: In value[[3L]](cond) : Invalid character in sequence. Cannot compute distance.
...

The rcpp method doesn't work either, but I suspect it's not the C code, but rather the character issue:

> dist_ham <- distToNearest(db, model="ham", rcpp=T)

 Rerun with Debug
 Error: incompatible sizes (1 != 4) 
4 stop(structure(list(message = "incompatible sizes (1 != 4)", 
    call = NULL, cppstack = NULL), .Names = c("message", "call", 
"cppstack"), class = c("Rcpp::exception", "C++Error", "error", 
"condition"))) 
3 rbind_all(x, .id) 
2 dplyr::bind_rows(list_db) at DistToNearest.R#691
1 distToNearest(db, model = "ham", rcpp = T) 

Same issue reported on Mac OS X by a user.

[Jason Vander Heiden]

The error may not be character related at all, as that messages is just the error message of the tryCatch around getSeqDistance.

[Jason Vander Heiden]

The error message from the tryCatch is misleading... This was an argument mismatch error, caused by a shazam/alakazam version mismatch (no rcpp option in the non-dev version of alakazam).

We should find a better way to deal with errors.

[Namita Gupta]

At least temporary fix of warning being the actual error with NA returned.