I would say remove all the /interface ethernet switch config, and try the basic common configuration with bridge vlan filtering activated:
/interface bridge vlan
add bridge=bridge_szkielet vlan-ids=30 tagged=bridge_szkielet,4_omni
add bridge=bridge_szkielet vlan-ids=200 tagged=bridge_szkielet,4_omni
/interface bridge port
set [find interface=interface=5_rejestrator_GI_R01] pvid=30
/interface bridge set bridge_szkielet vlan-filtering=yes
The above should configure the switch chip silently as well, and the forwarding and tagging/untagging operations should still be done in hardware.
But if routing between the subnet in VLAN 30 and the subnet in VLAN 200 is done by the CPU of the CRS112, the high CPU load is caused by this, not by bridging in software. Your OP is a bit inconsistent - you wrote that the computer in VLAN 200 has IP 192.168.5.2/24 and the VCR in VLAN 30 has IP 192.168.2.200, but in the config export, the VLAN subinterfaces of bridge_szkielet have a 10.10. prefix instead of 192.168. If this is a half-done obfuscation, and the default gateway of the VCR is 10.10.2.1 (or 192.168.2.1) and the default gateway of the computer is 10.10.5.254 (or 192.168.5.254), this is the case. The CPU of this device is sufficient to route its own traffic, not the traffic of connected devices.
Don't write novels, post /export hide-sensitive file=x. Use find&replace in your favourite text editor to systematically replace all occurrences of each public IP address potentially identifying you by a distinctive pattern such as my.public.ip.1.