Some years ago I did research on sealing aluminum tracks to concrete. What I found out is that Silicone is the sealant of choice for this task, hands down. A properly applied silicone seal will move + or - 100% in compression, tension, and shear! Note that this is 100% of the thickness of the sealant. That means that if the chainplate passes tightly through the deck then the sealant thickness will be minimal to nonexistant. It also means that ANY motion of the chainplate will break the seal! There are boats out there with chainplates that do not move I would guess, but not that most of us can afford.
If you can't keep the chainplate from moving then the obvious answer is to make the sealant thick enough to allow the motion without breaking. So simply opening up the deck all around the chainplate to 1/4" or 3/8" and filling the gap with silicone will allow the chainplate to move up, down or sideways 1/4" to 3/8", and never break the seal! To keep silicone from leaking below while it cures you simply stuff the bottom of the gap with the foam backer rod made for this purpose..
Feel free to sue silicone on your boat but having worked in boat yards and having owned over 20 boats of my own I can assure you silicone is exactly the worst product to use for chain plates for exactly some of the reasons you point out such as movement.
Silicone permeates the gelcoat and permanently pollutes & contaminates it so that nothing will stick to it in the future, not even 5200 successfully. Short of grinding and sanding the surface and silicone residue away there is virtually nothing you can do. It may look clean, and you've even washed it with acetone, but hmmmm it's still beading water??
There are also no commercially available chemicals to safely strip silicone from fiberglass or gelcoat without causing damage. If there was such a product someone would be very rich as my boat yard alone would buy two 55 gallon drums in a heart beat!! The ill effects and future problems associated with silicones use is documented in reports all over the sailing/web world and by those of use who have extensively worked on boats where owners used silicone.
There are very few good uses for silicone on boats due to it's profuse contamination of the surface and it's low adhesion properties. It also does not like to stick to stainless steel so in a chain plate it really is an overall poor choice.
Better choices for chain plates would include:
Sikaflex 291 = 220psi or 3M 4200/UV4000 = 300psi or 3M 101 or Life Caulk = Both about 140 PSI adhesion
I personally prefer 3M 101 due to it's natural UV resistance and ease of removal for future re-beddings. It also sticks to stainless steel very well when compared to silicone. My second choice, and one I've used many times, is Sikaflex 291. For a one part poly it has surprisingly good UV resistance and a good bond to stailnless.
Here's a quick run down:
3M UV 4000 is a one part polyurethane with UV inhibitors added. It has a elongation before break of 800% and a tensile strength of 300 PSI.
3M 4200 is a one part polyurethane with NO UV inhibitors added. It has a elongation before break of 900% and a tensile strength of 300 PSI.
3M UV 4000 is basically, spec wise, 4200 with UV inhibitors added which reduces is elongation before break slightly. This should NOT be a concern as 300 PSI is much more than necessary anyway.
3M 101 is a one part POLYSULFIDE! Polysulfides have a natural UV resistance and are the premier product for use in teak decks where UV rays bash them day in and day out. Polyurethanes do not do well with UV so UV inhibitors are added. 3M 101 has a tensile strength of 139 PSI and an elongation before break of 416%. Polysulfides are not compatible with some plastics.
Sikaflex 291 is also a one part polyurethane with a preferred lower adhesion than UV 4000 or 4200.. It offers an elongation at break of roughly 600% and is the easiest one part polyurethane sealant to get off in the future at roughly 221 PSI!
3M Marine Silicone offers the lowest adhesion of any 3M sealant and also the LOWEST elongation before break of any of the marine rated 3M sealants. 3M 4200 and 3M UV 4000 for example offer nearly double the flexibility before separation/break than does 3M Marine Silicone. 3M Silicone also has a very weak bond to metal and is only suggested for bedding or caulking metals by 3M not "adhering" to them as you actually need in a chain plate. 4200 and 4000 get an "H" rating on metal which mean they have a "High" strength bond to stainless. If the joint is moving, as chain plates do, you want a product that offers not only good elongation before break but also a decent level of adhesion to both the metal and the fiberglass. You also want one that won't have contaminated the surface for the next round of re-bedding to stick to.
So what's the bottom line? Both polysulfides & polyurethanes offer more elongation before break and adhere better to stainless than does silicone and they won't contaminate the surface for future re-beddings..
The advice about making the gap bigger around the chain plate is spot on! I bevel mine at about a 45 degree angle to get a nice thick gasket. If the deck is cored arond the chain plates remove them, dig out the core and back fill with thickened epoxy! On many production boat the chain plates are bolted to "screwed in" bulkheads, a poor & corner cutting design choice IMHO, and they WILL move!!
Again it's your boat and all I can do is present my findings & years and years of seeing silicone wreak havoc on boat decks and cores...
The only thing I would add would be to slightly countersink or bevel those screw holes to create an o-ring effect of sealant. I would have also dug into the area around the chain plates and made them bigger and really cleaned the stainless chain plates al the way through the deck to get a good bond of the new 3M 101..
This hole has been countersunk and will create a nice thick o-ring of sealant around the machine screw head: