So, in this case the client always will use the translated IP.
When VPN subnet translation is configured, the MX will check the source IP address against a address translation table. When 192.168.128.44 attempts to send traffic to the web server across the VPN, the source IP address is evaluated to be contained within the local subnet of 192.168.128.0/24, which requires a translation to be performed. The MX will then map the client's IP to the equivalent IP in the translated subnet. When the example client's traffic egresses the site-to-site VPN, it will have an IP address of 10.15.30.44.
If VPN subnet translation is configured, the translated subnet will automatically be advertised to all remote site-to-site VPN participants.
Full doc: https://documentation.meraki.com/MX/Site-to-site_VPN/Using_Site-to-site_VPN_Translation
I am not a Cisco Meraki employee. My suggestions are based on documentation of Meraki best practices and day-to-day experience.
Please, if this post was useful, leave your kudos and mark it as solved.