fix: minor fixes to some informational findings

[lekhovitsky]

Fixes:

• https://github.com/spearbit-audits/review-gearbox/issues/30
• https://github.com/spearbit-audits/review-gearbox/issues/35
• https://github.com/spearbit-audits/review-gearbox/issues/45
• https://github.com/spearbit-audits/review-gearbox/issues/67
• https://github.com/spearbit-audits/review-gearbox/issues/108

[StErMi]

@lekhovitsky there's a typo here: redice → reduce

[StErMi]

@lekhovitsky I think that the usage of the USE_SAFE_PRICES_FLAG flag (see issue https://github.com/spearbit-audits/review-gearbox/issues/108) should also be documented in the CreditFacadeV3 itself and not only on the PriceOracleV3.

The PriceOracleV3 is where the safe price behavior is configured, but it's the CreditFacadeV3 contract that will tell the oracle to use them or not.

Another issue about the safe price: with the PR #268 now also the "adapter logic" of the multicall could enable the USE_SAFE_PRICES_FLAG flag. This scenario is not explicitly documented in the PR and should be.

TLDR: Gearbox should:

• Document the usage of the "Safe Price" mechanism also in the CreditFacadeV3 contract
• Add to both the documentation of "Safe Price" in the CreditFacadeV3 and PriceOracleV3 contracts, the scenario enabling the USE_SAFE_PRICES_FLAG flag because of an external call to the adapter in multicall

[StErMi]

The code looks good and can be merged once the issues described in the above comments are integrated in this PR.

[lekhovitsky]

@StErMi Fixed a typo + added a comment about additional checks in _multicall's NatSpec in the latest commit