Here's how you can do this with BEDOPS bedops --partition and bedmap --count, if you have that toolkit available on your HPC or can install it via the Bioconda channel.

Start by creating a sorted BED file from your dataframe text file:

tail -n+2 CNVs.txt | sort-bed - > CNVs.bed 

Then partition the CNVs and map the CNVs against the partitions, count-ing the number of CNVs that fall over a partition:

bedmap --echo --count --delim '\t' <(bedops --partition CNVs.bed) CNVs.bed > answer.bed

If you just want the last three columns, i.e., to strip out the chromosome (not recommended):

cut -f2- answer.bed > answer.txt

I have developed this

 lines = [(200,1000),
         (400,800),
         (600,1500),
        ]


all_numbers = [(x[0],"O") for x in lines] + [(x[1],"C") for x in lines]

all_numbers = sorted(all_numbers, key=lambda x: x[0])


now_open = 0

for first_tuple, second_tuple in  zip(all_numbers, all_numbers[1:]):

    begin, what_side = first_tuple
    end, _ = second_tuple

    if what_side == "O":
        now_open += 1
    elif what_side == "C":
        now_open -= 1


    print(begin, end, now_open)

With Python ... don't need any python library.