Added support for digits param for each column

[RomanTsegelskyi]

As requested in #146, added support to format using separate digits param for each column.

> t
               drat    wt  qsec vs
Mazda RX4      3.90 2.620 16.46  0
Mazda RX4 Wag  3.90 2.875 17.02  0
Datsun 710     3.85 2.320 18.61  1
Hornet 4 Drive 3.08 3.215 19.44  1
> pandoc.table(t, digits = c(0, 10, 0, 0))

---------------------------------------------
                 drat   wt    qsec   vs 
-------------------- ------ ----- ------ ----
   **Mazda RX4**       4    2.62    16    0  

 **Mazda RX4 Wag**     4    2.875   17    0  

   **Datsun 710**      4    2.32    19    1  

 **Hornet 4 Drive**    3    3.215   19    1  
---------------------------------------------

> pandoc.table(t, digits = c(0, 10, 2, 0))

---------------------------------------------
                 drat   wt    qsec   vs 
-------------------- ------ ----- ------ ----
   **Mazda RX4**       4    2.62  16.46   0  

 **Mazda RX4 Wag**     4    2.875 17.02   0  

   **Datsun 710**      4    2.32  18.61   1  

 **Hornet 4 Drive**    3    3.215 19.44   1  
---------------------------------------------

[daroczig]

@RomanTsegelskyi are you sure you did not mix up digits and the round parameters?

[RomanTsegelskyi]

@daroczig, I am sorry, got confused with original request in #146 where digits param specifies number of decimal digits vs. what is currently supported by pander (number of total digits enforced by format).

> mt
               drat    wt  qsec vs
Mazda RX4      3.90 2.620 16.46  0
Mazda RX4 Wag  3.90 2.875 17.02  0
Datsun 710     3.85 2.320 18.61  1
Hornet 4 Drive 3.08 3.215 19.44  1
> pander(mt, digits = 3)

--------------------------------------------
                 drat   wt   qsec   vs 
-------------------- ------ ---- ------ ----
   **Mazda RX4**      3.9   2.62  16.5   0  

 **Mazda RX4 Wag**    3.9   2.88   17    0  

   **Datsun 710**     3.85  2.32  18.6   1  

 **Hornet 4 Drive**   3.08  3.21  19.4   1  
--------------------------------------------

> xtable(mt, digits = 3)
% latex table generated in R 3.1.3 by xtable 1.7-4 package
% Tue Jun 30 21:57:31 2015
\begin{table}[ht]
\centering
\begin{tabular}{rrrrr}
  \hline
 & drat & wt & qsec & vs \\ 
  \hline
Mazda RX4 & 3.900 & 2.620 & 16.460 & 0.000 \\ 
  Mazda RX4 Wag & 3.900 & 2.875 & 17.020 & 0.000 \\ 
  Datsun 710 & 3.850 & 2.320 & 18.610 & 1.000 \\ 
  Hornet 4 Drive & 3.080 & 3.215 & 19.440 & 1.000 \\ 
   \hline
\end{tabular}
\end{table}

By looking online seems that former is harder to achieve (SO), but IMHO makes a bit more sense than the latter. I was almost successful getting similar same behaviour as we currently have just with vector for digits, like

> pander(mt, digits = c(1,4,2,0))

---------------------------------------------
                 drat   wt    qsec   vs 
-------------------- ------ ----- ------ ----
   **Mazda RX4**       4    2.62    16    0  

 **Mazda RX4 Wag**     4    2.875   17    0  

   **Datsun 710**      4    2.32    19    1  

 **Hornet 4 Drive**    3    3.215   19    1  
---------------------------------------------

However, I started wondering if we should keep the current approach or switch to one like in xtable?

[daroczig]

I'd stick with the separate digits and round parameters -- although it might not be trivial for a newcomer, but that's pretty standard in R, and a nice approach anyway.

I'd go with an approach where both digits and round parameters can be provided as a vector. What do you think?

[RomanTsegelskyi]

I think that is a reasonable suggestion. I fixed the implementation accordingly. It came out a bit more complicated that I expected, due to need to handle matrixes separately because matrix columns can't be formated separately (as soon as first column is formatted all others are formatted too).

[RomanTsegelskyi]

Before only empty 2D objects were converted to data.frame, all other were untouched. Now conversion to data.frame will also happen for anything that inherits matrix, but just for digits/round. I checked an it doesn't seem to cause any additional complications.

[daroczig]

Looks good, thanks.