[devin-ai-integration]

Sets optimizer = true (and spells via_ir) so that the build matches the deployed implementations again: since forge 1.0 the optimizer is off by default.

[devin-ai-integration]

Does the optimizer setting change the deployed bytecode? Yes: with the current foundry.toml, forge ≥ 1.0 builds without the optimizer and the result no longer matches what is deployed.

I built MorphoTokenEthereum and MorphoTokenOptimism from main with forge 1.8.7 and compared them with the deployed implementations, ignoring the metadata:

	Ethereum 0x4364fd2371b6318159366abfa51f190df5c24852	Base 0xa860498f8a299526174b539fcc49f13cc082fb18
optimizer off (main)	15768 bytes vs 14162 deployed	16390 bytes vs 15175 deployed
optimizer on (this PR)	same size; only the UUPS self-address immutable differs	same size; only immutables differ (UUPS self address, bridge, remoteToken)

So this PR restores a build reproducing the deployment.