That's a little bit hard question, because any language is so big that very often we can find extremely rare, but still possible exception.
Generally, I would say in a phrase only one は can be thematic. The idea of the topic is to disambiguate. Any sentence has content which we want to say and some area to which it belongs. For example, if we provide only content like "will go to a shop", then such sentence doesn't make any sense, because we don't know about who we are talking here. Such sentence is possible with context as:
- And what you will do?
- Go to a shop.
It's not very common in English, but still understandable. We have context-topic "you/I" and action "go to a shop". It's very hard to think about any situation when we would need to use double layer disambiguation and the closest version is probably a very standard sentence like:
像は鼻が長い (elephants have long trunks)
像の鼻 は・が 長い (Elephants's trunks are long)
These sentences have different nuances due to different topics, but it's easy to see when we need to talk about some part of a whole, we can use の and string as far as we want. We can bring some general topic and narrow down to something much more specific within a single は. At the same time such disambiguation function gives another usage. If we intentionally emphasize there are several candidates and our sentence is true for one of these, people can assume it's false for others and thus we get contrastive implication. it's similar to a situation like this:
- What about the Smiths?
- John will come (which implies others members of the family probably won't do so).
With the only difference here we used context to make plural candidates, but in Japanese it's done simply by using は. It's the main function of は to restrict and If we use は, we automatically mean it could be applied to something else. Thus you can see so called contrastive function is rather an intention of speaker to emphasize on a choice and there are many ways to make that. In can be a flow of conversation when we start to bring similar topics, or it can be intentional usage of は in unusual places. We can make double particles like には and it's a place where there is no need for は in a neutral sentence, we can replace other particles like をー＞は as in ケーキは食べる (I will eat the cake), which implies person probably won't eat something else, or in some rare cases we can even put は in relative clause to make contrast.
At the same time there are many words like 実は (truth, reality) or situations like negation ではない, which technically require は particle. Everything we say can be either truth or lie. And while it's rather contrastive, many people use that as a filler word similar to "honestly speaking". Without intention to emphasize, that doesn't bring significant counter implication. It's also a little bit hard to categorize, because the only difference between two is that thematic-は is an old information. Any sentence has new information (something we want to say) and old information (something about what we say). Thus there is only one thematic-は and all other forms like 実は bring something new. But that's a little bit arguable and depending on how we try to explain what は does, that can be either another thematic-は, contrastive-は or maybe an exception. We mix here old/new information, disambiguation and implication. All 3 are slightly different concepts.
In my opinion both statements are rather true, because there is a huge gap between technically contrastive words/forms, and intentional implication done by speakers. But picking more precise definition would be something like "There is only one thematic-は, which can be contrastive or not depending on context, and any other は is contrastive by default with a different amount of implication".
