RCPP - optimize use of quantile / tdigest

[rspadim]

> td=tdigest(x)
> microbenchmark(
+     tdigest::tdigest(x),
+     tdigest::tquantile(td, 0.4),
+     quantile(x, 0.4),
+     sort(x)
+ )
Unit: microseconds
                        expr       min         lq        mean    median         uq        max neval
         tdigest::tdigest(x) 81299.500 90134.1015 98235.96202 94811.951 98986.0005 272509.501   100
 tdigest::tquantile(td, 0.4)     5.702    11.5015    41.84706    31.551    42.4005    236.202   100
            quantile(x, 0.4) 12648.402 14265.2510 17013.55897 15825.701 17950.6510  28711.101   100
                     sort(x)  8111.701  9770.2010 12486.58689 11088.801 13189.7005  37959.101   100

[OVVO-Financial]

quantile was used because initializing or creating a t-digest structure is computationally expensive.