What this PR does / why we need it:

The specific bug is with how field iterators are built around existing segments and used for in-memory index segment compaction. There was an initial Next call to each segment's field iterator, causing the iterator to start at e1 instead of e0. This is masked/uncommon because of how the iterator sorts these fields, and how the default field doc.IDReservedFieldName (i.e. "_m3ninx_id") alphanumerically precedes most fields and isn't required for field/term indexing. However, fields that start with a double underscore that precedes the ID reserved field name, e.g. "__m2" becomes susceptible for being skipped.

In practice for Prometheus metrics this has not been an issue since fields with double underscores are meant to be used only for internal use as per guidance:

Label names beginning with __ are reserved for internal use.

https://github.com/prometheus/docs/blob/e7567ffa4079b745bf0b4b5c8ef87401d2330588/content/docs/concepts/data_model.md?plain=1#L34-L36

Special notes for your reviewer:

Does this PR introduce a user-facing and/or backwards incompatible change?:

NONE

Does this PR require updating code package or user-facing documentation?:

NONE