Wide shots, you use a locked down tripod. You can do multiple takes, and layer together to create the illusion of many henchmen. When you do close ups, you won't see as many, so you can do close ups of a pair fighting, and put another pair or two in the background, to create the illusion of being in the middle of the crowed. In your fight area, you need to mark positions, so you can remember where everyone is. Doing this, you can do diffefent wide shots, from various angles. Then when doing close ups and mid shots, you can follow your ground markings to know who and how many other fighters might fall into frame.
What he meant be creating th camera shake. On your lock down shots obviously the camera will be stationary. You can use a shaky cam plug in, to give some camera shake to your lock down shots. So it's not too jarring to the viewer, going from locked off shot, to shaky handheld action shot.
If you plan this right, you could use 3 pairs of fighters, and create the illusion of dozens of fighters.
You have to plan, plan, plan, plan, and plan some more. Draw some diagrams, couple stick figure storyboards. You cant wing it on a shot like this.