Workflow Performance Bottleneck and Output Discrepancy in R UDF Operator + Scatter Plot Alpha Adjustment Request

Hi team,

I'm Ha from Steve Parker lab at U of Michigan. I'm testing local node Texera with this workflow and I found a few issues.

1. The workflow takes quite a while to get it done. This takes around 20 minutes. When using python, it only takes 5 minutes. I think sorting and R UDF operator are the bottlenecks, esp. with sorting tuples. Attached files are the workflow and raw data files with ~550k lines.

[hpap-019-metrics-csv.zip](https://github.com/user-attachments/files/20847856/hpap-019-metrics-csv.zip)

[barcode_rank_plot.json](https://github.com/user-attachments/files/20847853/barcode_rank_plot.json)

2. The number of output tuples shown on the R UDF operator should be the same as the number of input tuples, but the numbers are different.


3. Can we add an option to change the `alpha` value of the scatter plot?

This is the result of the python code I used with the `alpha` value being 0.2.

This is the result of the texera workflow. It is using a default `alpha` value.

Start by reproducing the reported workflow with the attached raw data and barcode_rank_plot.json, then investigate the R UDF output count and scatter plot alpha behavior. The issue also reports a sorting and R UDF performance gap, so completion would require defining and addressing the relevant performance, output-count, and visualization discrepancies.

Written by the indexing model from the issue text.