RAW - No, Otto's Irresistible Dance is not a Charm spell
It does not say that the target will be "charmed", and the condition it imposes is significantly different than the description of the "Charmed" condition.
OID:
A dancing creature must use all its movement to dance without leaving its space and has disadvantage on Dexterity saving throws and attack rolls. While the target is affected by this spell, other creatures have advantage on attack rolls against it.
Charmed:
A charmed creature can't attack the charmer or target the charmer with harmful abilities or magical effects. The charmer has advantage on any ability check to interact socially with the creature.
A dancing creature can attack the caster (with disadvantage -- and that disadvantage applies to any attack, not just against the caster) and can cast AoE spells that include the caster and anyone else they choose in the target area with no penalty. They just can't move freely. A charmed character can move freely but can't attack or target the caster.
Furthermore, if it were a Charm spell, it would not be necessary to specify that "[c]reatures that can't be charmed are immune to this spell." That appears to be listed as a special exception.
RAI - No, Otto's Irresistible Dance does not impose the Charmed condition
Jeremy Crawford has tweeted saying:
Being charmed means being subjected to the charmed condition
Since OID does not impose the Charmed condition, the Fey Ancestry resistance to "being charmed" does not apply.
What makes sense?
Webster's dictionary says:
charm: to affect by or as if by magic : compel
Magically forcing someone to dance seems to fall completely within that definition.
I would give those with Fey Ancestry advantage on saves against all enchantment spells that compel the user to do something that is not their own free choice. This includes many spells that do not specifically say they are "charms" or that they impose the "Charmed" condition, such as command, compelled duel, and yes, Otto's irresistible dance. I believe those all fall within the common sense and dictionary definitions of "charmed" in this context, and there's nothing in the description of Fey Ancestry that says it only applies to things that grant the "Charmed" condition.
Yes, if the second Goblin is not within the Reach of the thing making the opportunity attack
It seems a bit weird, however the criteria for triggering an opportunity attack are:
- You are moving out of the reach of a hostile creature
- This movement is from one of your Actions, Bonus Actions, Movement or Reaction
- You have not taken the Disengage action
- You are not Teleporting
For this question we can discount Point 4, as the Goblin Boss can't teleport.
We'll also ignore Disengaging for now.
For Point 2, the movement is by the Goblin Boss' Reaction, which fulfills that criteria
For Point 1, if the swapping with the second Goblin does not cause the Goblin Boss to leave the Reach of any hostile creature, it will not trigger an Opportunity attack. In the diagrams below, the first set of movement will not trigger an opportunity attack, while the second set will.
EG -> EB
B G
EBG -> EGB
E - Enemy;
G - Goblin;
B - Goblin Boss
Note that wherever the second Goblin is moved from, it does not trigger an Opportunity attack from anything as its movement is forced by the Goblin Boss.
As for using Disengage (via Nimble Escape), it only stops OAs for the rest of your Turn, not the entire Round.
So, if someone had Readied an action to hit the Goblin Boss when the Boss did a specific thing, then the Goblin Boss did that thing on the Boss' turn, and in Reaction the Boss used Redirect Attack, that would not trigger an OA, wherever it moved, as it is still the Boss' Turn so Disengage is still in effect. But normally, Disengage would not help for this.
Best Answer
You are right
The text says the following
The text doesn't mention that "another goblin" has to be an ally of the Goblin Boss, then it would say "another allied goblin" or something similar, and it also doesn't say that the goblin has to be opposed to the creature attacking the Goblin Boss.
Therefore the Goblin Boss can also choose a PC Goblin that happens to be close by or an NPC that is opposed to the Goblin Boss.
As a DM I wouldn't do this because it feels to me like this is not the intention behind the ability. I have always imagined this to be "grab an unsuspecting ally and hold him in front of me as a shield", which is already difficult, but grabbing an enemy or hiding behind them is a lot harder because they are actively fighting you.
Also, as nitsua60 mentioned in the comments, the MM was published roughly 3 years before VGtM, which first allowed PCs to choose Goblins as a race. This means that the fact that players could play Goblins was likely not taken into account - because at that point there was no such option - which means that they didn't have to think about this when wording the ability and none of the playtesters could have come up with this usage inside the framework of the existing official rules.
Additional information from Jeremy Crawford, mentioned by Rubiksmoose: