caution about different versions of doRedis on master and workers (if on network)

[trcook]

This is not as much a bug as it is an advisement (that might be useful to indicate in change-logs or the documentation): the doredis workers and the doredis master need to be running the same version of doRedis if they are not on the same machine.

I've found this after following advice listed in #24 and installing the latest version of doRedis to my master node (but not my worker nodes, which are baked into images on a remote machine)

If the setup is:

host1:
|  master----------------------->|
|  (dev-version of doRedis 1.2.0)|
|                                |
|                                |
|  |<---------redis-server <---- |
|  |
|  |-------->localWorker
|  (last stable CRAN doRedis 1.1.1)

Then everything works out fine.

if, however, the setup has redis and the workers on a different machine, Then, the differences in version number cause and issue
I.e. when the setup looks like this

host-1:
|  master----------------------->|
|  (dev-version of doRedis 1.2.0)|
                                 |
host-2 :                         |
|                                |
|  |<---------redis-server <---- |
|  |
|  |-------->localWorker
|  (last stable CRAN doRedis 1.1.1)

Then the error returned will suggest some sort of issue with the way the worker reads and processes the jobs:

## Waiting for doRedis jobs.
## Processing job  Error in cat(list(...), file, sep, fill, labels, append) :
##  argument 2 (type 'list') cannot be handled by 'cat'
##  In addition: Warning messages:
## 1: In sprintf("%s:%.0f.env", queue, ID) : NAs introduced by coercion
## 2: In sprintf("%s:%.0f.results", queue, ID) : NAs introduced by coercion

If the workers are running the latest (dev) version of doRedis (1.2.0), then this works as expected (which is quite marvelous and has great potential for using doRedis for scaled HPC tasks : ) .