1-2. If you are talking about a single one-to-one match, it doesn't make any sense to talk about "overlapping" segments. You have segments in common and their location (based on some matching threshold) or none at all. It only makes sense to talk about "overlapping" segments when you are comparing multiple matches to one another for "shared segments."
In this case, a "shared segment" that is long enough normally will point to a single common ancestor (not a pair; although there may be a pair, and if a pair only had surviving children with one another, it may only be possible to associate a segment to pair rather than an individual).
What "long enough" means is subject to some wiggle room. Often 7 cM is considered significant. But bear in mind that especially given variables such as the number of SNPs, it's not set in stone. A "7 cM" segment could actually be a 3 cM segment from one shared ancestor and a 4 cM segment from another shared ancestor. Or an SNP-sparse, identical-by-state segments.
2-3. About the only conclusion one might draw is that the individual with the longer shared segment inherited more DNA from the common ancestor. The match with the longer segment might be from an older generation. That's not necessarily the case especially when you're talking about a single segment. Powerful evidence of the degree of relationship comes from comparing total cM count with many segments.
4. The best two links I can give you are these:
- Autosomal DNA Statistics
- Shared CM Project
120 cM is by far most consistent with "Second cousins once removed, half second cousins, first cousin three times removed, half first cousin twice removed" etc., but it's possible for it to be a degree closer or further than that.
5. Can you clarify which service you are using? Is this Family Tree DNA? How are you sorting people by ancestral line?