[BUG] github workflows: build-linux-all jobs running out of disk space

[kv2019i]

Describe the bug
The VMs used for build-linux have started to run out of space again, causing test jobs to fail systematically with recent SOF main:

https://github.com/thesofproject/sof/actions/runs/20127912805/job/57761888425?pr=10429

xtensa-nxp_imx_adsp_zephyr-elf/bin/xtensa-nxp_imx_adsp_zephyr-elf-ranlib modules/sof/libmodules_sof.a && :
/github/home/zephyr-sdk-0.17.4/xtensa-nxp_imx_adsp_zephyr-elf/bin/xtensa-nxp_imx_adsp_zephyr-elf-ar: unable to copy file 'modules/sof/libmodules_sof.a'; reason: Success

[kv2019i]

@cgturner1 can you take look at? I remember we did one round of pruning already at one point that bought us a bit more time (last time):

[cgturner1]

yea i'll see what i can squeeze out again. Last time we slimmed the zephyr docker to include only the relevant toolchains maybe I can try a little bit of slimming down in the workflow itself but this is getting tricker to do as we are limited on the free github runners SOF uses

[cgturner1]

PR opened to de-duplicate the build and only build unique targets from --all step #10434