Reduce memory usage of skip reader

### Description

Following up on #13576. Currently, we always use the size `Lucene99PostingsFormat.MAX_SKIP_LEVELS` to initialize the skip reader.

https://github.com/apache/lucene/blob/30c965ea575a5e75c2bf724a340aa690d82f1ec5/lucene/core/src/java/org/apache/lucene/codecs/lucene99/Lucene99PostingsReader.java#L496-L501

We can take the max docs into consideration to calculate the maximum number of skip levels like we did in `MultiLevelSkipListWriter`.

https://github.com/apache/lucene/blob/30c965ea575a5e75c2bf724a340aa690d82f1ec5/lucene/core/src/java/org/apache/lucene/codecs/MultiLevelSkipListWriter.java#L75C8-L82

Start in Lucene99PostingsReader.java around the skip-reader initialization linked in the issue, then compare the maximum-level calculation in MultiLevelSkipListWriter.java. Confirm that the reader's allocation accounts for the maximum document count rather than always using Lucene99PostingsFormat.MAX_SKIP_LEVELS, and verify the relevant Lucene tests.

Written by the indexing model from the issue text.