addTA creates legend (when legend=NULL) when plotting on an existing subchart

For example:

``` r
library(quantmod)
data(sample_matrix)
x <- as.xts(sample_matrix)
chartSeries(x, theme=chartTheme("white"))
addTA(Cl(x)*1.01, on=1, legend=NULL) # adds legend even though legend=NULL
addTA(Cl(x), legend=NULL) # no legend (new subplot)
addTA(Op(x), on=2, legend=NULL, col="green") # adds legend
```


The only legend that should be on the above image is "Last 47.76...". The relevant line(s) of code seem to be in the `chartTA` function in `R/TA.R`, specifically:

``` r
if(!x@new) {
legend <- function(legend,text.col,...) {
list(legend=legend,text.col=text.col)
}
}
```

But I'm not sure what needs to change to fix this issue without breaking anything.

Start with the reproduction using chartSeries and addTA, then inspect chartTA in R/TA.R, especially the !x@new legend handling. Trace how legend=NULL is treated for existing versus new subcharts. Done means the reproduction creates no legend for the calls with legend=NULL while preserving the expected "Last 47.76..." legend.

Written by the indexing model from the issue text.