My understanding is that when sequencing a genome, "gaps" refer to fragments which were not sequencing in the sequencer.
Is this correctly? What are there other meanings to the term "gap"?
You're not wrong but gap can mean more than that. A good recent reference is this paper: Genetic variation and the de novo assembly of human genomes.
Essentially, a gap occurs if something happens in our genome that can't be explained by uniformity. Since our genome is highly complicated, you can expect gap is also more than just mis-sequencing. The gap you mention is the
sequence-coverage gaps mentioned below.
This paper has a section about the types of gaps.
1. Sequence-coverage gaps
Sequencing gaps occur, under the simplest condition, where no sequence reads have been sampled for a particular portion of the genome.
2. Segmental duplication-associated gaps
Over one-third (206/540) of the euchromatic gaps in the human reference genome (GRCh38) are flanked by large, highly identical segmental duplications.
3. Satellite-associated gaps
These include short and long runs of tandem repeats designated as short tandem repeats (STRs; also known as microsatellites), variable number of tandem repeats (VNTRs; also known as macrosatellites) and Mb-sized centromeric satellite repeats.
4. Muted gaps
Muted gaps are defined as regions that are inadvertently closed in an assembly but that actually show additional or different sequences in the vast majority of individuals
5. Allelic variation gaps
Some regions of a genome also show extraordinary patterns of allelic variation, often reflecting deep evolutionary coalescence.
The word ‘gap’ can be used both loosely and in specific genomic contexts, as listed by @Student T. The usage that seems to relate to your question — Sequence Coverage Gaps — is quite nicely illustrated in the Wikipedia entry on ‘Contig’. I reproduce, below, a modified form of a (public domain) illustration from it.
The sequences of the ‘contigs’ (see legend) are known completely as the ‘reads’ (solid blue arrows in legend) overlap, even though the ‘fragments’ (see legend) contain unsequenced regions. However because the length of the fragments can be determined roughly (e.g. by gel electrophoresis) — here fragment 5 is the key — the two contigs can be assembled into a ‘scaffold’ with a gap of (approx.) known length. I have marked this as ‘Gap’ in green.