APPS load_key_certs_crls(): Make file access errors much more readable

[DDvO]

Since the use of OSSL_STORE for file loading in apps, the output in case of file access errors has become needlessly bulky and hard to read, for instance:

apps/openssl x509 -in WRONG
Could not open file or uri for loading certificate from WRONG
00517F71887F0000:error:16000069:STORE routines:ossl_store_get0_loader_int:unregistered scheme:crypto/store/store_register.c:237:scheme=file
00517F71887F0000:error:80000002:system library:file_open:No such file or directory:providers/implementations/storemgmt/file_store.c:269:calling stat(WRONG)
Unable to load certificate

It turns out that the first error queue entry is spurious, so this PR prunes it.
The second entry is quite informative, but not nice to read for app users,
so I've made sure that the errno string is printed directly instead.
This way, the output on the above app invocation becomes

Could not open file or uri for loading certificate from WRONG: No such file or directory

[DDvO]

Doing the latter improvement in apps/lib/app.c, I found that #15670 had introduced a pretty ugly way of suppressing DH parameter loading errors.
This PR changes this to a much less invasive way of doing this: temporarily unset the bio_err variable.

[t8m]

Doing the latter improvement in apps/lib/app.c, I found that #15670 had introduced a pretty ugly way of suppressing DH parameter loading errors.
This PR changes this to a much less invasive way of doing this: temporarily unset the bio_err variable.

I have to disagree that your way is less ugly. 😁

[t8m]

IMO this is not something that should be merged before 3.0 release.

[DDvO]

Doing the latter improvement in apps/lib/app.c, I found that #15670 had introduced a pretty ugly way of suppressing DH parameter loading errors.
This PR changes this to a much less invasive way of doing this: temporarily unset the bio_err variable.

I have to disagree that your way is less ugly. grin

Why? It takes much less code than adding a wrapper function and making the various print function calls explicitly dependent on an explicit suppress_decode_errors parameter.

[DDvO]

IMO this is not something that should be merged before 3.0 release.

I advocate for merging this before the release because it corrects issues introduced with OSSL_STORE in 3.0.

[DDvO]

Also handling of this PR should be resumed, since we are after the 3.0 release.

[paulidale]

Just one style nit.

[DDvO]

Merged - thanks @paulidale