While I also think this is the likeliest scenario, I wouldn't exclude one where PLANs emphasis on anti-shipping capabilities as you said resulted in a similar route as the Soviet Navy fielding the Echo, Charlie and Oscar SSGNs for this specific reason.
I see it mainly as a budgetary decision.
The US is ramping production to 4 SSN-equivalents per year, where an SSBN is roughly two SSNs.
I wouldn't be surprised if China is aiming for 6+ SSN-equivalents per year, given:
1. the size of the new Huludao facilities
2. the USA becoming an aggressive and
unpredictable imperial power, particularly in the past year
3. the production rates elsewhere with the Chinese Navy (at or reaching twice the US level in other warship categories)
4. China being significantly behind in quiet nuclear submarine numbers, compared to other warship categories.
---
At 6 per year, this could be:
4x Type-095 SSN
1x SSBN/SSGN
---
A notional large SSGN design (based on the SSBN) would be significantly more efficient in terms of [platform cost per missile tube] when compared to a stretched Type-095.
And when you already have 4 SSNs being produced per year, there is space in the force structure for a large specialised SSGN.