You've got the definitions mostly right, but the distinction usually comes down to who holds the routing table (BGP) and who pays whom.
Here is how we usually define these in the field:
1. ISP Allocated Range (PA Space)
If i subscribed to an ISP ... is it consider IP transit right?
No, this is usually just "Dedicated Internet Access" (DIA). In this scenario, you are a "stub" network. You don't have an ASN, and you aren't running BGP (usually). You are just a customer inside the ISP's network. They handle all the global routing for you. You are just pointing a default route to them.
2. Your ASN + Your IPs + Upstreams
is this still consider IP transit?
Yes, strictly speaking. If you have your own ASN and you are announcing your prefixes to two providers who carry your traffic to the rest of the world, you are buying IP Transit. You are "transiting" their network to reach the rest of the internet. This is a standard "Multi-homed" setup.
3. Your ASN + IX Connection
this will then be consider IP peering?
Yes. Connecting to an Internet Exchange (IX) and setting up BGP sessions with other members (like Google, Akamai, or other ISPs) is Public Peering.
As to your last question: "Why would one choose (3) over (2)?"
You generally don't choose Peering instead of Transit. You use Peering to reduce your Transit.
If you only have Peering (Option 3), you can only talk to the other networks on that specific exchange. You cannot reach the rest of the random websites on the internet unless one of those peers sells you transit.
You add Peering to your mix because:
Cost: Peering at an IX is often cheaper per Mbps than Transit.
Performance: If you push a lot of traffic to a specific network (e.g., you are an ISP and your users watch a lot of Netflix), peering directly with Netflix at an IX is faster than sending that traffic up through your Transit provider and back down.
Summary: You need Transit (2) to reach the whole world (Global Table). You add Peering (3) to offload heavy traffic and save money/latency.