What is the purpose of using label_replace when it's discarded by means of a `without` operator?

I'm following this article here and there's a part where the author adds a label, but this label is immediately discarded by the surrounding query function. Here is the code:

- record: job:http_requests:rate5m_prediction
  expr: >
   quantile(0.5,
     label_replace(
       avg_over_time(job:http_requests:rate5m[4h] offset 166h)
       + job:http_requests:rate5m:avg_over_time_1w - job:http_requests:rate5m:avg_over_time_1w offset 1w
       , "offset", "1w", "", "")
     or
     label_replace(
       avg_over_time(job:http_requests:rate5m[4h] offset 334h)
       + job:http_requests:rate5m:avg_over_time_1w - job:http_requests:rate5m:avg_over_time_1w offset 2w
       , "offset", "2w", "", "")
     or
     label_replace(
       avg_over_time(job:http_requests:rate5m[4h] offset 502h)
       + job:http_requests:rate5m:avg_over_time_1w - job:http_requests:rate5m:avg_over_time_1w offset 3w
       , "offset", "3w", "", "")
   )
   without (offset)

According to my understanding, the instant vectors are going to be concatenated by the or operator, and then their labels are going to be combined, only for the offset label to be discarded (by the without operator). Wouldn't it be the same to omit the label_replace function?