Which platform data structure, security, etc is better equipped to
support a pan-region decentralized identity platform with more than
280 million people?
Sadly, none! As of Nov 2018, we haven't even reached 100M created wallets worldwide and we barely have 100 monthly active users on dApps.
As identity is concerned, "better" and "self-sovereign" are esoteric, relative terms. The elephant in the room is the fact that Internet identities imply completely different mental models compared to what peoples are accustomed to: plastic cards issued by nation states. Thus, forget all you know about identity and how you think about it, because there's hardly any way to port the traditional models onto blockchains.
ERC725 and ERC735 rely on the wisdom of the crowd. To use them, you assume fragmented trust across the network. If a doctor you know, your neighbour and a reputable business all pledge that address
$x is indeed "John Smith", then you could trust that "John Smith" owns address
$x - however, your dApp could mandate any number of claims required to deem a person trustworthy.
Because these ERCs are, in theory, implemented as smart contracts on top of Ethereum (a borderless network), it will be, to say the least, non-trivial for nation states to casually migrate their databases. There are dozens of companies like Chainalysis who run BigData operations, being able to find out what's happening on public blockchains and who owns what.
Note: privacy-enforcing features such as zkSNARKs or zkSTARKs might be a massive game changer here, but, as of Nov 2018, they are either (1) not implemented at all in Ethereum or (2) prohibitively expensive and slow to compute.
Hyperledger Indy has a different approach insofar it is a standalone RBFT blockchain specifically built for identity purposes. This means weaker assumptions and it might be an effective solution for governments to step in. The procedures for validating identity are analogous to ERC725, i.e. they provide ways for people or organisations to vouch for others and so forth.
Naturally, because you don't have to run the smart contract code on all the blockchain nodes, you have enhanced technical capabilities, but, at the same, there's trust required and transparency is optional (whether the code is open source or not).
Zooko's triangle states that an ideal naming system would be human-meaningful, secure and decentralised, but this isn't possible to achieve (sidenote: it actually is, blockchains solved this, have a look on ENS or Namecoin).
This is just an opinionated belief, but I find the term "self-sovereign" highly misleading. Cryptocurrency, as a general concept, represents self-sovereign money because, no matter what happens to the big bros, you can flee the country and preserve your wealth. Conversely, if a large organisation becomes corrupt, that might have a ripple effect across many parties in these meta-identity platforms.
If you're indeed self-sovereign, shouldn't you be able to always change your name without input from other entities? Update your height, weight, hair colour, relationship status etc? We're just a quirky collection of atoms. There's a long way ahead before we manage to fully sync atoms and bits.