Sidechains and hardforks are definitely not the same thing.
As outlined in the original Why Cardano essay, the whole point of Layer 1 Cardano is to serve as a settlement layer for domain-specific computation performed by a collection of Layer 2 "sidechains". The explicit bifurcation of one settlement layer and many execution layers allows for much of the heavy computation to be spread out with domain-specificity.
The extent to which a sidechain inherits Cardano's security depends on how frequently and tightly they interact. Although there may be overlap, sidechains can and will have their own accounting models, consensus mechanisms, validator sets, e.t.c. Their "security" will depend on which of the many trade offs they've made.
So, the decision of whether to sidechain or hardfork depends on what you are trying to achieve. Are you trying to deploy some killer new execution/consensus/dApp engine? A privacy chain? Do you want smooth interoperability with everyone else's new chain? Go sidechain.
Or, do you want to make such radical changes to the settlement layer of all this infrastructure that the HFC and voltaire-era governance procedures are not enough? Then hardfork. (This is unlikely to be viable).